How to read the map

Release 124 EV charging freshness boundary

The optional layer remains 27,922 reviewed BEE facilities dated 26 October 2025, now filterable by exact operator and connector fields. A 21 July 2026 official aggregate reports 52,718 stations, including 16,561 with fast chargers for cars, but has no row crosswalk. It adds no points or facility-level operation claims.

Release 123 cold-chain boundary

Release 123 is the validated current data release. The Ministry's complete inventory dated 31 March 2026 supplies 408 source-listed approved cold-chain project entries with sector, district, State/UT, approval date, typed money quantities, and evidence-dated Physical Progress.

• Lifecycle: Completed is not proof of commissioning, inauguration, present operation, or 100% progress. Ongoing is not proof of physical construction, delay, stalling, or present activity after the cutoff.
• Money: project cost, revised approved grant-in-aid, and grant released remain distinct. Twenty-four released-grant cells remain missing; the row-level revised-grant sum conflicts with the printed grand total, and two rows contain internally impossible ordering.
• Organizations: the inventory does not establish owner, sponsor, operator, contractor, concessionaire, or EPC roles. The Ministry remains labelled only as reporting ministry.
• Geography: markers are separately attributed district, locality, or broad State/UT representative points in WGS84. They are navigation aids, not project sites, facility footprints, service areas, or geometry.

Release 122 Mega Food Parks boundary

Release 122 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The Ministry's complete table dated 31 August 2025 supplies 41 project identities, final-approval dates, project-cost values, approved and released grants, source-listed SPVs, and evidence-dated statuses.

• Lifecycle: table wording remains an observation at the source cutoff. Operational and Complete becomes two dated events but neither is carried forward as a present-day claim.
• Money: project cost, approved grant, and released grant remain distinct. The approved-grant row sum is ₹1,950.384 crore while the printed total is ₹1,956.834 crore; the ₹6.45 crore source difference is not silently repaired.
• Organizations: the named SPV keeps the exact source-grounded role Source-listed SPV / implementing agency. Contractor, owner, operator, concessionaire, and EPC roles remain unpublished unless a separate source establishes them.
• Geography: all markers are separately attributed representative locality points in WGS84. They are navigation aids, not surveyed sites, parcels, facilities, gates, or project boundaries.

Release 121 non-major-port project boundary

Release 121 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Appendix III of the Ministry's April to September 2025 ports-sector update publishes 21 ongoing private-sector, captive or joint-venture non-major-port project rows with a status cutoff of 30 September 2025. Eighteen rows meet the baseline inventory contract.

One second Honnavar row is linked to the same canonical PPP port-development project while retaining its conflicting capacity, cost and status wording. Redi and Subarnarekha remain needs-review candidates because no unambiguous project anchor was established. An unrelated locality or river point is not substituted.

Project cost and capacity retain the table's exact labels and limitations. Karanja's printed 0.0 cost remains unresolved rather than becoming a zero-cost claim. Operational, construction, clearance and agreement wording is evidence-dated, not carried forward as present status. Every accepted marker is representative navigation, not port limits, berth geometry or a surveyed project site.

Release 120 current NIXI directory boundary

Release 120 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. NIXI's public NOC directory and its row-level API expose 79 current listings, exactly matching the official 18 June 2026 aggregate. Forty-two rows link to stable Release 119 records and thirty-seven create new baseline Digital & Telecom facilities.

A current directory listing is not a lifecycle state. The API publication flag is not interpreted as operation, commissioning, uptime, capacity, traffic, ownership, hosting tenure, or service quality. Only the 42 linked records retain their separate reported-operational event dated 9 December 2022; the other 37 publish no inferred row-level operation.

The directory's address text remains attributable evidence, but its embedded maps are excluded because the review found wrong-city, displaced, and missing references. Each marker uses a separately attributed GeoNames locality navigation point and remains representative, never an exchange rack, building footprint, route, or service boundary. Shared locality points remain unmoved.

Contact names, phone numbers, and email addresses visible in the public source remain only in the frozen audit artifact and are omitted from product records.

Release 119 NIXI Internet Exchange boundary

Release 119 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Rajya Sabha Unstarred Question 404 names 42 NIXI Internet Exchanges reported installed and operational on 9 December 2022. Each source row becomes one baseline Digital & Telecom facility record.

The lifecycle wording remains bounded to the 2022 evidence date. A later official statement dated 18 June 2026 reports 79 IXPs nationwide but does not publish the location list. The unlisted 37 are therefore an explicit coverage gap, not inferred records or proof about the current state of any named 2022 facility.

The official annexure supplies place and State/UT but no facility coordinate or geometry. Every marker is separately attributed representative locality navigation. Chennai 1 and Chennai 2 retain one shared Chennai anchor without displacement; Chandivali uses broader Mumbai orientation. No point represents an exchange site, rack, data centre, route, service boundary, or surveyed geometry.

Release 116 CEA renewable-project boundary

Release 116 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The sixth bounded partition reviews numbered rows 145 through 155 from the frozen March 2026 CEA report. Ten become new canonical projects and the blank-name Rajasthan-only row remains an auditable needs-review candidate.

Capacity for execution, commissioned capacity, and under-construction capacity remain separate source fields. Compound, qualified, and NA schedule cells stay literal unless both compared columns contain one valid compatible date.

Every accepted marker uses separately attributed representative navigation. The Nokh and Phalodi district difference, exact no-match responses, broader Bikaner orientation, and shared Khambhaliya point remain explicit. None becomes a project site, turbine, solar array, parcel, route, boundary, or surveyed geometry.

Release 115 CEA renewable-project boundary

Release 115 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The fifth bounded partition reviews numbered rows 105 through 144 from the frozen March 2026 CEA report. Eighteen become new canonical projects and twenty-two remain auditable needs-review candidates.

Capacity for execution, commissioned capacity, and under-construction capacity remain separate source fields. The register does not establish physical-progress percentage, present output, current post-cutoff operation, cost, expenditure, exact project geometry, or causation for schedule movement.

Every accepted marker uses separately attributed representative navigation. Exact no-match results and broader administrative fallbacks stay explicit. State-only rows, repeated developer or holding-company identities, uncertain revisions or components, and the unresolved Hunsagi/Bijapur placement remain held rather than being repaired through inference.

Release 114 CEA renewable-project boundary

Release 114 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The fourth bounded partition reviews numbered rows 82 through 104 from the frozen March 2026 CEA report. Fifteen become new canonical projects and eight remain auditable needs-review candidates.

Capacity for execution, commissioned capacity, and under-construction capacity remain separate source fields. The register does not establish physical-progress percentage, present output, current post-cutoff operation, cost, expenditure, exact project geometry, or causation for schedule movement.

All accepted markers use separately attributed representative navigation. Non-finalized names, zero-active-capacity rows, the multi-state Rajasthan and Karnataka row, and the unclear portfolio identity remain held rather than being repaired through inference. Tentative schedule qualifiers remain visible.

Release 113 CEA renewable-project boundary

Release 113 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The third bounded partition reviews numbered rows 62 through 81 from the frozen March 2026 CEA report. Twelve become new canonical projects, three link to existing records, and five remain auditable needs-review candidates.

Execution, commissioned, and under-construction capacity remain separate source fields. The register does not establish a physical-progress percentage, present output, current post-cutoff operation, cost, expenditure, exact project geometry, or causation for schedule movement.

All accepted markers use separately attributed representative navigation. The held rows preserve an internally inconsistent capacity split, the Bhadla and Bikampur location conflict, and three unconfirmed relationships to DVC Floating Solar Park Phase I instead of resolving them through inference.

Release 112 CEA renewable-project boundary

Release 112 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The complete March 2026 CEA report remains frozen once. This checkpoint reviews numbered rows 31 through 61 only: twelve become new canonical projects, five exact named systems link to earlier solar-park records, and fourteen remain auditable needs-review candidates.

Capacity for execution, commissioned capacity, and under-construction capacity remain literal source columns. Scheduled and anticipated completion dates retain their separate meanings. None establishes present progress, output, storage, expenditure, completion, commissioning, or current operation.

CEA publishes no project coordinates or geometry in this table. The twelve points use separately attributed representative places or landmarks. Konar, Raghunathpur, and Mejia landmarks are orientation context only; shared and administrative anchors remain unmoved. The Bagodara row preserves the source difference between its project name and the Surendranagar location column.

Mirzapur and Odisha source capacities conflict with earlier canonical system records. The new rows remain linked observations without overwriting the prior values or creating duplicate projects. Repeated names, multi-state or distributed scope, State-only geography, and the row carrying the literal source type 10 remain held rather than being mapped through inference.

Release 111 CEA renewable-project boundary

Release 111 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The complete March 2026 CEA under-construction renewable-project report is frozen, while this checkpoint reviews only numbered NTPC rows 1 through 30. Twenty-seven become new projects, Barethi and NTPC REL Khavda link to existing canonical solar-park systems, and the Maharashtra-only row remains needs-review.

Capacity for execution, commissioned capacity, and under-construction capacity remain distinct source columns. They are not physical progress, current output, storage capacity, expenditure, or proof that every nested component is operating. The source role remains Source-listed developer.

Scheduled and anticipated completion are compared only when both dates are valid and chronologically compatible. The source's invalid 31 June 2026 Gadag date and the Fatehgarh-I anticipated date that precedes its scheduled date remain explicit conflicts without a derived movement.

CEA publishes no project coordinates or geometry in this table. All twenty-seven markers use separately attributed representative navigation. Shared and multi-site anchors are not jittered and never become project sites, turbines, arrays, parcels, routes, boundaries, or surveyed geometry.

The separately audited April 2026 CEA BESS report contains national aggregates but no project identities. The linked May 2026 PDF returned 404 during review. No BESS project record was created from aggregate totals or a broken source.

Release 110 CEA solar-park system boundary

Release 110 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The frozen March 2026 CEA report contains 55 numbered Solar Park / UMREPP rows. Fifty-two pass the baseline identity and geography contract; Patoda, Bodana, and UP Solar Park remain auditable needs-review candidates and create no marker.

Total park capacity remains a park-system quantity, not commissioned generation, output, present operation, or storage capacity. The mixed Actual/Anticipated Solar Park Completion Date column is interpreted only when its wording supports the distinction. A completion entry does not establish every nested plant, transmission component, land parcel, contract, commissioning event, or present operation.

The source-published agency role remains Source-listed solar park agency. It is not silently converted into owner, contractor, operator, concessionaire, funder, or regulator.

The PDF publishes place descriptions but no reusable project coordinates or park boundaries. Separately attributed locality, landmark, and administrative points remain representative navigation with explicit limitations and no jitter.

Release 109 CEA hydro execution boundary

Release 109 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The frozen June 2026 CEA under-implementation workbook contains 40 rows: 29 conventional hydroelectric projects and 11 pumped-storage projects. Twenty-one rows pass as new canonical projects, while 19 link to existing records and do not count twice.

Under implementation remains a source-dated state as of 30 June 2026. The workbook's Date of finish/commissioning remains a current reported target from that source. Neither field establishes present construction activity, original or revised deadline movement, completion, inauguration, commissioning, opening, operation, delay, or performance after the cutoff.

Each named organization retains the exact source role Executing agency. No owner, sponsor, funder, contractor, concessionaire, operator, project company, award, expenditure, payment, current cost, actual cost, or final cost is inferred from the workbook.

The workbook publishes no reusable project coordinate or complete geometry. All 21 new markers use separately attributed representative locality navigation. They are not CEA project coordinates, dam sites, powerhouses, reservoir extents, tunnel alignments, parcels, surveyed worksites, or complete project geometry. Shared Sileru locality anchors remain shared without jitter.

Release 108 LPAI operational land-port boundary

Release 108 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The current LPAI roster and two June 2026 Ministry releases reconcile to fifteen named operational land ports. Only those fifteen publish. The separately mentioned eleven planned facilities remain outside the candidate universe because the reviewed source does not name them.

The Ministry's 7 June 2026 statement establishes a source-dated operational state and the exact LPAI operating and managing role. It does not establish opening, completion, commissioning, inauguration, uninterrupted later operation, traffic, capacity, condition, performance, contractor, concessionaire, customs operator, ownership, cost, expenditure, award, payment, or audit evidence.

Each map point is a separately attributed OpenStreetMap/Nominatim town, village, or locality anchor. Sutarkandi uses disclosed Karimganj orientation context and Srimantapur disclosed Sonamura context because their official pages describe those facilities by distance from those places. No point is presented as a gate, terminal, customs building, parcel, facility boundary, surveyed coordinate, or complete geometry.

Sabroom keeps a visible source difference: the newer Ministry statement places it in the current operational roster, while an older LPAI detail page retains September 2021 foundation-stone and pre-construction wording. The newer dated status governs the current read without deleting the historical wording.

Release 107 IBM Gandhinagar inspection-register boundary

Release 107 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Three frozen Indian Bureau of Mines Gandhinagar 2025-26 register pages preserve 120 candidate rows and 25 linked mine-specific reports. Seven unique mine codes pass stable identity, register/report consistency, and conservative representative-locality review, bringing the local catalog to 10,287 records and the default non-toll scope to 10,198.

The lifecycle label MCDR inspection recorded describes a dated regulatory event only. It does not establish present operation, production, compliance, completion, commissioning, suspension, closure, reserves, safety, or performance. Five register/report date conflicts, missing or invalid stable codes, unavailable reports, and unresolved geography remain held.

Four accepted points use separately attributed locality records. Three use clearly labelled public locality landmarks where the reviewed navigation address establishes the exact source village, district, and state. Every point remains representative. It is not a mine site, lease centroid, extraction footprint, surveyed boundary, or project geometry. Source-published Devanagari names remain intact, while deterministic English search aliases do not replace official identity.

Release 106 IBM inspection-register boundary

Release 106 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Five frozen Indian Bureau of Mines 2025-26 regional MCDR registers preserve 180 candidate rows and 170 linked mine-specific reports. Twenty-four unique mine codes pass stable identity, register/report consistency, and conservative representative-locality review, bringing the local catalog to 10,280 records and the default non-toll scope to 10,191.

The lifecycle label MCDR inspection recorded describes a dated evidence event only. It does not establish present operation, production, compliance, completion, commissioning, suspension, reserves, safety, performance, or ownership beyond the exact published role. A violation or show-cause letter listed by IBM is an attention signal, not an adjudication or finding of misconduct, corruption, blame, or project failure.

The reviewed report fields do not publish exact mine coordinates or lease geometry. Each accepted map point is separately attributed OpenStreetMap/Nominatim locality navigation matching the report's locality, district, and state. It is not a mine site, lease centroid, extraction footprint, surveyed boundary, or project geometry. Personal contact and inspection-administration fields visible in the public reports are not extracted into the product.

Release 105 space-launch evidence boundary

Release 105 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Seven accepted records expand the local candidate to 10,256 records: one spaceport, three launch pads, one sounding-rocket facility, one launch-complex development project, and one private launch facility. The entity kinds remain separate, and public source evidence is not extended into sensitive range geometry, security boundaries, unpublished facilities, or present mission readiness.

First Launch Pad, Second Launch Pad, and TERLS use published official facility coordinates. SDSC SHAR, Kulasekarapattinam Launch Complex, Third Launch Pad, and Agnikul Launchpad-01 use separately attributable representative anchors. Active, operational, inaugurated, approved, under construction, first launch, completion, commissioning, opening, and continuous present operation remain separate meanings.

Four explicit child relationships connect the First, Second, and Third Launch Pads and Agnikul Launchpad-01 to SDSC SHAR without merging records or moving their stored coordinates. Skyroot's Infinity Campus and Max-Q Campus remain needs review because the reviewed official sources do not yet reconcile the identities and relationship strongly enough for canonical publication.

Release 104 shipbuilding and ship-repair evidence boundary

Release 104 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Seven accepted records expand the local candidate to 10,249 records: two shipbuilding or ship-repair development projects, two shipyard assets, and three ship-repair facilities. The entity kinds remain separate and the lane does not turn vessel contracts, individual ships, historical facility lists, or planned developments into accepted infrastructure records.

The Kochi New Dry Dock and the International Ship Repair Facility preserve the 17 January 2024 inauguration event separately from FY2024-25 reported-operation evidence. Hooghly Cochin Shipyard's source-published dedication to the nation remains a milestone, not an inauguration, completion, commissioning, opening, or operation inference. Dated operating statements do not prove present operation after their reporting cutoff.

Every accepted record uses an official address or facility description plus a separately attributed representative navigation point. None is a shipyard boundary, dry-dock outline, berth, parcel, production area, worksite, security extent, or surveyed facility geometry. The source point, CRS, retrieval date, precision class, derivation, and limitation remain attributable.

Differently dated ₹168 crore, ₹180 crore, and ₹175.20 crore observations for Hooghly Cochin Shipyard remain non-comparable, and the source wording for the ₹70 crore Udupi facility value remains exact. The interface derives no cost revision from these quantities. Organization labels retain exact company, parent-company, lessor, MoU-counterparty, and operating roles.

The 2025 Kochi block-fabrication and Tamil Nadu greenfield-shipyard proposals remain needs review because the reviewed sources do not establish a publishable approved site and source-backed position. No authoritative reusable source was located for the reported Tata Kerala facility, so it remains outside the candidate ledger. Naval, strategic, sensitive, and inferred geometry remain out of scope.

Release 103 EV charging read boundary

Release 103 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. The browser no longer downloads the complete EV facility catalog. The versioned server route requests only the current result page, exact matching count, and a zoom-aware set of individual source-locked points. Compatible D1 is the primary read; binding absence, version mismatch, or query failure activates a separately labelled server-side static fallback.

The D1 and static paths must agree on accepted-only scope, search, State/UT, district, viewport, direct-record selection, cursor order, and held-record exclusion. A viewport limits which individual points are sent for drawing; it does not change the exact matching count, move a coordinate, cluster records, or claim that every match is simultaneously visible. The selected brief retains the same BEE evidence and present-operation limits.

Local measurements are implementation evidence, not production claims. The default EV opt-in fallback response was roughly 221 KB, a named two-result search roughly 19 KB, and a dense close-view response roughly 507 KB. Production D1 latency, caching, and migration behavior remain unverified until the original owning account returns and the queued migrations pass a clean replay and full release gate.

Release 102 EV charging discovery boundary

Release 102 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. EV charging is a secondary facility layer and is off on a fresh map. Explicit activation, a deliberate matching search, or a direct facility URL loads the compact reviewed facility index and adds its records to the active count, results, map, and URL state. Home returns to the clean non-EV scope.

Each accepted record preserves BEE's published point, CPO column, connector rows, source Government or Private classification, and evidence cutoff through 26 October 2025. The mapped point is an official facility point from that dated source, not a surveyed charger-bay position, parcel, access guarantee, or proof of present operation. Missing live availability, hours, pricing, ownership, commissioning, and connector serviceability remain missing.

The default app does not request the EV artifact. The explicit layer uses individual native map features with exact counts and no clustering or coordinate movement. The current compressed static index is a local fallback and transition path; D1 viewport queries remain the production-scale target after the owning account returns and parity is revalidated.

Release 101 BEE public EV charging evidence boundary

Release 101 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Public production remains Release 97. The frozen Bureau of Energy Efficiency PDF contains 39,641 table rows through 26 October 2025. Exact-row deduplication and conservative source-identity grouping produce 29,366 facility candidates. BEE publishes no stable station ID, so candidate IDs are deterministic local review keys rather than official identifiers.

Independent district containment accepts 27,922 baseline facilities across all 36 represented states and union territories. Another 1,444 candidates remain needs review because of invalid coordinates, shared source coordinates, source-state conflicts, multiple coordinates for one normalized identity, no containing district, or unreadable PDF text. The map does not correct, jitter, geocode, or guess any held position.

The source supports an evidence-dated listing, CPO name, Government or Private source classification, place fields, coordinates, and published connector details. Inclusion does not prove present operation, live availability, public access hours, price, ownership, commissioning, inauguration, continuous service, or connector serviceability. The Government or Private column is not relabelled as ownership.

EV charging remains a secondary facility layer under the existing Logistics & Inland Transport category with a source-backed presentation subtype. These facilities contribute zero to the core 10,000-project count and do not enter the initial map. The current checkpoint stages the reviewed universe in local static and D1 artifacts only; interface search, filter, marker, result, URL, and selected-record integration remain pending.

Eight local SQL artifacts preserve every accepted and held disposition in a dedicated candidate table. A clean SQLite replay and second application retain the same 29,366 candidate rows, 27,922 accepted rows, 1,444 held rows, and 27,922 search rows. Remote D1 remains unchanged.

Release 100 Sleemanabad linked-component evidence boundary

Release 100 is LOCAL ONLY. It is not saved, deployed, or applied to remote D1. Public production remains Release 97. The reviewed Sleemanabad carrier-canal tunnel is linked to existing Bargi Diversion Phase III because its published RD 104–129 contract scope is fully inside the parent's km 104–154 scope. It does not create a second canonical project or marker.

The CAG report records a December 2007 tender, ₹799 crore turnkey award to Patel-SEW (JV), a 26 March 2008 work order with a 40-month tenure, a seventh extension, and work still in progress at the 30 June 2023 audit cutoff. It also records a separate ₹9.76 crore grout-block contract with Phaloudi Constructions and three audit observations. Department replies and CAG responses remain paired with every observation.

The ₹1,356.05 crore value is the up-to-date work value reported in the 338th Running Bill. It is not silently relabelled as total contract payment, expenditure, revised project cost, actual cost, or final cost. The two contract values and three audit-observation amounts remain separately typed and are not presented as a compatible cost-revision sequence.

The current parent-project snapshot remains 81.52% progress as of the June 2026 reporting cutoff, with a December 2026 revised completion target. Historical excavation, expert review, monitoring, extension, and under-execution evidence does not establish tunnel breakthrough, completion, lining completion, hydraulic testing, commissioning, water release, or operation. July 2026 breakthrough reports remain held because no attributable government release or stock-exchange filing was located.

The map preserves the parent's existing representative corridor point. No reviewed source supplies an attributable Sleemanabad alignment, tunnel portal, worksite, parcel, command-area polygon, or complete project geometry, so none is invented.

Release 99 Bhogapuram airport-development evidence boundary

Release 99 is a local candidate only. It is not saved, deployed, or applied to remote D1. Public production remains Release 97. The new record models Bhogapuram International Airport as an airport-development project, separate from the unresolved DGCA airport-asset candidate and the existing NHAI airport-connectivity package.

The Vizianagaram District Administration reports 98.74% construction progress as of 6 April 2026. The same source says aerodrome licensing, BCAS security clearance, Customs notification, and Immigration notification were in process. The record therefore does not claim completion, licence issuance, inauguration, commissioning, opening, commercial operation, or a fixed opening date.

The MoEFCC Form 1 publishes ₹2,461.27 crore as proposal project cost, the 2017 parliamentary answer gives an approximate ₹2,260 crore airport-construction estimate, and the 2023 PIB release gives ₹4,592 crore as the Phase 1 estimated project cost. Their dates, labels, and scopes differ, so the interface derives no cost-change amount or percentage. None is silently relabelled as award value, expenditure, payment, actual cost, or final cost.

GMR Visakhapatnam International Airport Limited retains the exact role Airport developer. Bhogapuram International Airport Corporation Limited remains the environmental-clearance applicant company, while Airports Authority of India and the Andhra Pradesh Government remain the two parties to the cited land-mutation and civil-enclave MoU. No EPC contractor, concession agreement, payment, audit, or broader organization responsibility is inferred.

The map normalizes the AAI degree-minute-second airport reference coordinate into WGS 84 display order and labels it as an official airport reference point. It is not a terminal, airport boundary, runway endpoint, parcel, construction footprint, or full project geometry. The public AAI PDF was reviewed, but repeated direct local downloads timed out; its URL and extracted coordinate remain attributable while no immutable local PDF hash is claimed.

Release 98 PPPAC Port project evidence boundary

Release 98 is a local candidate only. It is not saved, deployed, or applied to remote D1. Public production remains Release 97. The candidate audits all 43 Port rows in the official PPPAC project list: 37 new canonical projects, five linked rows that do not double-count an existing or revised project identity, and one 2009 Ennore row held for identity review.

Final approval and in-principle approval remain evidence-dated PPPAC appraisal outcomes. They do not establish a current concession, selected bidder, contract, construction state, completion, commissioning, opening, operation, ownership, or benefit. Ministry of Ports, Shipping and Waterways retains only the exact source-table role PPPAC table sponsoring authority.

VOC Outer Harbour and JNPT Fourth Container Terminal contain explicitly comparable appraised total-project-cost observations in their reviewed meeting records. Their calculated changes remain separate from other list values, grants, awards, expenditure, payments, actual cost, and final cost. No other row receives a cost-change claim without compatible published quantities.

The map reuses unchanged source-backed major-port references and separately frozen Diamond Harbour and Sahibganj navigation points. These remain official-facility or representative points, not project sites, port limits, terminal boundaries, berths, parcels, corridors, routes, or complete project extents. The 37 local additions pass coordinate-order, India-bounds, source State/UT, precision, and limitation checks. The independent Release 97 state and district polygon snapshots remain frozen at 10,204 released records, so no new polygon-containment result is claimed.

Release 97 PPPAC Airport project evidence boundary

Release 97 adds the complete 10-row Airport subset of the official PPPAC project list and lifts the catalog to 10,204 records: 10,115 core non-toll records and 89 optional toll-plaza facilities. The projects remain separate from the eight linked DGCA airport assets and from differently scoped terminal-construction projects already in the catalog.

The 85th meeting supports six operation, management, and development proposals and a table of estimated concessionaire investment. The product preserves that exact money meaning. It is not original cost, revised cost, award value, expenditure, payment, actual cost, or final cost.

The project list displays `0.00` for two selected-area O&M proposals and two city-side development proposals. The 80th meeting says no capital investment was required for its two O&M proposals, while the 14th meeting publishes no compatible project-cost quantity. The public project-cost field therefore remains Not published rather than becoming a zero-cost claim.

Eight final-approval and two in-principle-approval outcomes retain their meeting dates. They do not establish a current procurement, selected bidder, contract, concessionaire, operator, implementation state, completion, inauguration, commissioning, opening, or operation. Airports Authority of India or Ministry of Civil Aviation retains only the source-table role PPPAC table sponsoring authority.

Each project uses the unchanged representative navigation point of its exactly matched DGCA airport asset and passes independent State and district containment at that point. It is not a PPPAC coordinate, selected-area boundary, city-side parcel, concession extent, terminal location, worksite, or surveyed project geometry.

Release 96 Smart Cities ICCC evidence boundary

Release 96 adds 99 citywide Integrated Command and Control Centre baseline systems and lifts the catalog to 10,194 records: 10,105 core non-toll records and 89 optional toll-plaza facilities. The Open Government Data Platform city list retains all 100 source candidates under the Government Open Data License - India. New Town Kolkata remains needs review because the reviewed GeoNames extract does not publish one unambiguous locality feature.

MoHUA reported that ICCCs had been established across all 100 Smart Cities as of 15 November 2024. The interface displays this as an evidence-dated reported operational state. It does not present the statement as confirmed current operation in July 2026 or as evidence of uninterrupted service, integration completeness, performance, capacity, commissioning scope, ownership, or a city-specific control-room site.

Each accepted record uses a separately attributed GeoNames city feature under CC BY 4.0. Hubballi-Dharwad uses the documented arithmetic midpoint of the two published locality points. These markers are representative city-navigation aids, not ICCC buildings, service areas, municipal boundaries, connected assets, control rooms, or surveyed geometry.

Ministry of Housing and Urban Affairs retains only the exact role Source reporting ministry. City SPVs, owners, operators, contractors, selected bidders, tenders, contracts, amendments, costs, payments, audits, and stronger geometry remain missing until city-specific authoritative evidence is reviewed.

Release 95 Zojila Tunnel evidence boundary

Release 95 adds one integrated Zojila Tunnel and connecting-road project and lifts the catalog to 10,095 records: 10,006 core non-toll records and 89 optional toll-plaza facilities. It resolves PAIMANA candidate 618412 but preserves that row's Andhra Pradesh state field as an attributable source conflict. Dedicated NHIDCL and PIB sources support discovery across the Jammu and Kashmir–Ladakh corridor.

The current NHIDCL row reports ₹6,808.69 crore TPC, 69% physical progress, 63% financial progress, an original 30 September 2026 completion and a likely 28 February 2028 completion. The reported dates yield a 516-day movement. Because the row publishes no separate observation date, the progress uses the 19 July 2026 retrieval cutoff and says so directly.

The ₹4,509.50 crore successful quoted EPC award value, ₹3,215.86 crore PAIMANA cumulative expenditure, and current TPC remain different money types. They are not treated as original and revised project costs, payments to one contractor, actual cost, or final cost. NHIDCL, MoRTH, and MEIL retain their exact implementing-authority, line-ministry, and EPC-contractor roles.

The map converts two approximate coordinates printed in the NHIDCL tender schedule to WGS 84 decimal display coordinates and derives their arithmetic midpoint for navigation. The original endpoint texts and converted coordinates remain stored. The midpoint is not a portal, alignment, worksite, parcel, surveyed midpoint, or full project geometry.

The 9 June 2026 main-tunnel breakthrough remains a construction milestone rather than evidence of completion, commissioning, opening, or operation. A signed EPC agreement, amendments, certified payment ledger, package expenditure split, disputes, audits, actual cost, final cost, and completed or operating state were not located and remain explicit gaps.

Release 94 Vadhavan Port evidence boundary

Release 94 adds Vadhavan Port as one distinct development project and lifts the catalog to 10,094 records: 10,005 core non-toll records and 89 optional toll-plaza facilities. It does not convert the project into an operating port asset, merge it with a separately modelled road package, or resolve the still-held Galathea Bay candidate.

The Cabinet's ₹76,220 crore figure is stored as the approved whole-project total including land acquisition. A later VPPL award separately records ₹1,648 crore excluding GST for the near-shore reclamation and shore-protection EPC package against its ₹1,770 crore excluding GST package estimate. The package comparison is not displayed as whole-project cost movement, and none of these values becomes expenditure, payment, actual cost, or final cost.

The project company and delivery roles remain source-literal: Vadhvan Port Project Limited is the project SPV, Jawaharlal Nehru Port Authority and Maharashtra Maritime Board hold 74% and 26% respectively, and ITD Cementation India Ltd. is the EPC contractor for the named package. No broader contractor role is inferred.

The map uses the reviewed environmental filing's project reference coordinate in WGS 84 display order. That coordinate is evidence-backed navigation near the project area, but it is not a port limit, cadastral parcel, construction footprint, access corridor, or surveyed centre. A coarser coordinate printed on the JNPA page remains recorded as a source difference rather than silently replacing the more precise official reference.

The latest reviewed stage evidence is the 31 December 2024 package award. Physical progress, a current whole-project deadline, signed contract, expenditure, payments, current operation, actual cost, and final cost were not located in the reviewed official sources and remain visibly missing.

Release 91 ten-thousand-project inventory checkpoint

The public catalog contains 10,093 accepted source-backed records. Of these, 10,004 core non-toll projects, packages, infrastructure assets, and source-defined systems meet the baseline-or-better inventory contract. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

Release 91 re-reviews 438 previously unpublished records from two existing official source families. It accepts 129 Amrit Bharat Station Scheme station-development records and 131 NHAI Balance For Award roads or packages. Another 176 candidates remain in needs review and two source rows resolve as duplicates rather than inflating the catalog.

Ministry of Railways and NHAI sources continue to establish project identity and their dated status wording. The frozen GeoNames India extract and a frozen 9,356-feature OpenStreetMap railway-station snapshot provide separately attributed representative navigation only. They do not establish a station worksite, highway alignment, package midpoint, surveyed geometry, present operation, award, contractor, cost, schedule, progress, or evidence tier.

Four accepted NHAI packages retain a source-backed relationship to a broader source-named parent scope because the published package names and non-overlapping chainages support that relationship. Ambiguous phases, broader overlaps, and duplicate-looking rows remain held. D1 preserves all 438 dispositions, two duplicate decisions, four package relationships, and each source fingerprint in two idempotent migration shards.

Release 90 AMRUT 2.0 locality-navigation expansion

The public catalog contains 9,833 accepted source-backed records. Of these, 9,744 core non-toll projects, packages, infrastructure assets, and source-defined systems count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

Release 90 reviews all 2,984 Water Supply or Sewerage/Septage projects still pending after Release 88. It accepts 1,681 records across 25 states and union territories only where the official ULB label has exactly one same-state GeoNames place feature under exact normalized primary-or-alternate-name matching. Another 1,303 eligible projects remain held because the match is ambiguous or absent. Parks and WB Rejuvenation remain pending because their source labels alone do not support a maintained category assignment.

The 1,099 matched official ULB pages resolve to 1,098 distinct GeoNames features because Barrackpore and North Barrackpore share one navigation feature. Their official project identities remain separate and the shared source coordinate is not moved or jittered. GeoNames is CC BY 4.0 navigation geography only, not Ministry project evidence, a project site, facility, service area, network, route, parcel, surveyed extent, status, cost, schedule, contractor, or evidence tier.

The portal's Project Cost remains the literal source-defined quantity, and an undated portal milestone remains an observed directory label rather than a current lifecycle claim. D1 stores the complete 7,612-row residual ledger plus normalized links to the official AMRUT page and the separate GeoNames feature for each new record.

Release 89 CEA generation navigation re-review

The public catalog contains 8,152 accepted source-backed records. Of these, 8,063 core non-toll projects, packages, infrastructure assets, and source-defined systems count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

Release 89 re-reviews 110 non-hydro rows that remained pending after the January 2024 CEA monthly-generation lane and that publish positive generation for that month. Thirty-eight rows pass one conservative navigation match in the independently verified source state; 72 remain needs review. A match must use one unique named OpenStreetMap power=plant feature, compatible fuel, a normalized score of at least 0.70, at least 0.15 separation from the next candidate, and no high-confidence existing-catalog match.

CEA remains the authoritative source for station identity, source-grouped fuel, capacity monitored by that table, and the January 2024 generation observation. OpenStreetMap contributes only a separately attributed ODbL navigation point. A node keeps its published point; a way or relation uses the representative center returned by Overpass. Neither representation is an official CEA coordinate, surveyed plant boundary, unit location, facility footprint, present-operation finding, current capacity, ownership, operator, or contractor evidence.

The 1,958-page CEA plant-wise renewable-capacity PDF was also audited but deferred: it is predominantly historical, provides no coordinates, produced unreliable extracted location fields, and did not offer a clearer reuse basis than the already frozen open-data lane. No row from that PDF enters Release 89.

Release 88 reviewed-anchor AMRUT 2.0 inventory

The public catalog contains 8,114 accepted source-backed records. Of these, 8,025 core non-toll projects, packages, infrastructure assets, and source-defined systems count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

Release 88 accepts a deterministic, state-balanced batch of 200 residual rows from the frozen 8,769-row AMRUT 2.0 directory: 100 Water Supply and 100 Sewerage or Septage projects across 18 states and union territories. Every row uses one of 112 ULB locality anchors already reviewed in Release 79. The residual control plane remains complete with 200 accepted and published dispositions and 7,412 pending dispositions.

The portal's Project Cost value remains source-published project cost. It is not relabelled as original estimate, sanctioned cost, tender value, award, expenditure, revised cost, payment, actual cost, or final cost. The portal milestone remains an undated directory observation and is not carried forward as current stage, progress, completion, commissioning, opening, or operation.

The Ministry directory publishes no project coordinate, route, site, footprint, or service-area geometry. Each record uses a separately attributed representative locality point for the source-published ULB. Shared ULB anchors remain shared without jitter. They do not establish a plant site, intake, reservoir, pipe, sewer, drain, parcel, municipal boundary, or surveyed extent.

D1 preserves the 200 baseline records, all 7,612 residual candidate dispositions, typed source observations, exact source links, and spatial limitations in three idempotent migration shards. Static discovery, selected-record details, and evidence remain the independent public fallback while D1 serving parity is staged.

Release 87 Smart Cities civic-utility inventory

The public catalog contains 7,914 accepted source-backed records. Of these, 7,825 core non-toll projects, packages, infrastructure assets, and source-defined systems count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The bounded source universe contains 448 rows from 28 resources published through the Smart Cities Data Portal under the Government Open Data License - India. Release 87 accepts 305 distinct civic-utility facilities after explicit resource classification, row-identity review, cross-catalog duplicate review, coordinate-order normalization, India bounds, city-context review, and independent state and district containment. Ninety-one rows remain needs review, five are duplicates, and 47 aggregate, equipment, observation, or classification-conflict rows are rejected or deferred.

The accepted batch contains 129 Power Grid & Storage facilities, 91 Environmental Utilities facilities, 83 Water System facilities, and two Urban Infrastructure fire facilities across Maharashtra, Sikkim, Karnataka, Chandigarh, Gujarat, Mizoram, Tamil Nadu, Uttar Pradesh, and Nagaland. Category assignment comes from the reviewed resource scope and row fields, not cost or a vague name.

A source row supports only a dated facility listing and any literal attributes it publishes. A published commissioning year remains a year-level observation; it is not converted into an exact date or proof of current operation. The product does not infer present service, continuous operation, current condition, ownership, operator, capacity, progress, completion, performance, contract, cost, or municipal responsibility.

The accepted coordinates are official facility points from the reviewed resource rows. Decimal, degree-minute-second, and reversed coordinate-label cases are normalized deterministically into WGS 84 longitude-latitude order while original values remain preserved. Every accepted point matches its declared state and exactly one district. Three rows whose published points fall in another state remain needs review; the map does not move them to fit the label.

An official facility point is not a surveyed parcel, building footprint, service area, network, pipe, cable, drain, municipal boundary, or complete asset extent. D1 retains all 448 candidate dispositions, 28 resource documents, five deduplication decisions, 305 baseline maturity rows, typed source-row observations, exact source links, and the same spatial limitations. Static discovery and evidence reads remain the independent public fallback.

Release 86 PIB / BharatNet State/UT systems inventory

The public catalog contains 7,609 accepted source-backed records. Of these, 7,520 core non-toll projects, packages, infrastructure assets, and source-defined systems count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The reviewed Press Information Bureau answer published on 1 April 2026 contains 34 State/UT rows in its BharatNet Annexure-I. Release 86 accepts all 34 as baseline Digital & Telecom system records after identity review, taxonomy review, cross-catalog deduplication, India bounds, and independent state and district checks. They are subordinate implementation scopes within the source-named national BharatNet Programme, not 34 separate national programmes or physical facilities.

The source reports 269,062 GPs or equivalents and 218,347 service-ready points. A service-ready point is not a unique Gram Panchayat, fibre route, tower, exchange, data centre, contract, completed project, commissioned network, operating connection, or surveyed feature. The source warns that one GP can have more than one service-ready point, so the product does not calculate a completion percentage or a performance verdict.

Department of Telecommunications remains the exact Programme implementing department. Ministry of Panchayati Raj remains the exact Source reporting ministry. Neither role is silently relabelled as owner, contractor, concessionaire, operator, funder, auditor, regulator, or package-delivery organization.

The annexure publishes no physical coordinate or reusable network geometry. Each record uses a separately attributable Survey of India cartographic State/UT label placement as an administrative navigation anchor. It is not a BharatNet site, route, facility, centroid, office, parcel, or surveyed extent. Four anchors sit outside the independent land polygons and Puducherry's anchor intersects Tamil Nadu; the source scope, coordinate, and discrepancy remain visible without moving or upgrading the marker.

Release 85 Ministry of Jal Shakti / PIB Namami Gange inventory

The public catalog contains 7,575 accepted source-backed records. Of these, 7,486 core non-toll projects, packages, and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The reviewed Ministry of Jal Shakti answer published by PIB on 7 February 2022 contains 41 named Namami Gange sewerage-project rows across Uttarakhand, Uttar Pradesh, Bihar, Jharkhand, and West Bengal. Release 85 accepts 36 baseline Environmental Utilities records after extraction, cross-catalog identity review, representative-locality review, India bounds, and independent state and district containment. Two duplicate rows link to existing canonical projects and three ambiguous locality candidates remain needs review.

Every project status and expenditure quantity has a 31 December 2021 cutoff. Completed, Completed. (Under O&M), Under Progress, tendering, and AA&ES wording remain dated source observations. They are not carried forward as current 2026 construction, completion, commissioning, opening, operation, tender, or contract status. Source completion wording remains historical wording rather than a confirmed current deadline or outcome.

Approved project cost remains approved project cost. Total expenditure incurred remains cumulative expenditure at the historical cutoff. The quantities stay separately typed and are not compared as original versus revised cost, award versus payment, actual cost, or final cost. STP capacity is capacity to be created, not achieved treatment or present utilization.

The annexure publishes no reusable project coordinate, site, alignment, footprint, parcel, or service-area geometry. Each accepted record uses a separately attributed OpenStreetMap locality representative. The point is navigation only and does not establish an STP site, drain or sewer alignment, treatment-plant footprint, project boundary, or surveyed position.

Release 84 Ministry of Railways / PIB project-status inventory

The public catalog contains 7,539 accepted source-backed records. Of these, 7,450 core non-toll projects, packages, and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The reviewed Ministry of Railways answer published by PIB on 11 March 2026 contains 163 unique named project rows across Bihar, Madhya Pradesh, Gujarat, and Uttar Pradesh. Release 84 accepts 116 baseline Railway records after table deduplication, cross-catalog identity review, named-endpoint navigation review, India bounds, and independent state and district containment. Thirty-three duplicate rows and 14 unresolved navigation candidates remain auditable and do not count again.

Completed means only that a row appears under the source's completed or recently completed heading. The source row does not publish a completion date, and the product does not infer commissioning, opening, present operation, 100% progress, or final cost. Taken up, Works taken up, and Recently sanctioned retain those exact source meanings and do not become tender award, construction start, or current status.

The table's Cost or Latest Cost quantity is stored as source-published project cost. It is not relabelled as original estimate, sanctioned cost, revised cost, expenditure, tender value, award value, payment, actual cost, or final cost. Project, commissioned, and balance lengths retain their source definitions and are not converted into map geometry or a synthetic whole-project progress percentage.

The source publishes no reusable coordinate or railway alignment. Each accepted record uses a separately attributed OpenStreetMap named endpoint or locality whose own returned address independently agrees with the reviewed source-state scope. The point remains representative navigation only and is not a Ministry coordinate, surveyed worksite, corridor midpoint, route, parcel, or project extent.

Release 83 NHAI Balance For Award inventory

The public catalog contains 7,423 accepted source-backed records. Of these, 7,334 core non-toll projects, packages, and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

NHAI's Balance For Award report contains 821 numbered rows as of 1 March 2026. Release 83 accepts 181 distinct road-development projects or packages after complete-table extraction, source-state parsing, a minimum source-defined length gate, exact and fuzzy catalog review, representative-navigation review, India bounds, and independent state and district containment.

Inclusion in Balance For Award establishes only that the row appeared in that dated report. It does not establish award, signed contract, construction start, current status, physical progress, completion, opening, operation, delay, performance, or final cost. Intended mode does not prove award. Total length is not completed length or geometry. The literal DPR Name cell is not relabelled as contractor, concessionaire, owner, authority, operator, or awarded consultant.

NHAI publishes no reusable coordinate, chainage geometry, alignment, route, package midpoint, site boundary, or surveyed extent in the reviewed report. Each accepted record uses a separately attributed OpenStreetMap place or administrative representative that matches the reviewed anchor and expected state context. The point remains navigation only.

The complete ledger remains auditable: 181 accepted, nine exact duplicate rows, 38 needs-review records, 224 pending-navigation records, and 369 rejected or deferred rows. Candidate, duplicate, unresolved, deferred, and pending rows do not count toward the catalog. D1 preserves every disposition, source fingerprint, source link, evidence observation, exact source-publisher role, spatial limitation, and deduplication decision.

Release 82 NICDC industrial-corridor inventory

The public catalog contains 7,242 accepted source-backed records. Of these, 7,153 core non-toll projects and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

NICDC's Delivery Monitoring Unit report dated 30 June 2026 contains 27 report-defined project candidates. Seventeen pass as new canonical Industrial Corridor records. Ten sections link to existing canonical projects and do not create another record, marker, or count. Parent-corridor narratives remain context rather than projects.

Every stage remains evidence-dated. Completed trunk-infrastructure packages do not prove whole-project completion, commissioning, occupancy, operation, utilization, or present condition. Dedication and foundation-stone events do not silently become completion or operation. Typed approved, planning, transferred, activation, identified, or proposed areas are not rendered as surveyed polygons.

Organization roles remain exact: Reporting corporation, Project SPV named in report, EPC contractor named in report, EPCC named in report, and Programme Manager for New Cities named in report. NICDC is not silently relabeled as project owner or authority, and approved packages, matching equity, debt, and contract awards are not collapsed into one project cost.

The report publishes no reusable coordinate or geometry. Sixteen records use separately attributed locality navigation points. The Jharkhand record uses the reviewed Survey of India state-label point and remains explicitly unresolved because the report says an alternate site is being identified. None of these points is presented as a project site, land boundary, corridor route, or surveyed extent.

Release 81 prioritised MMLP inventory

The public catalog contains 7,225 accepted source-backed records. Of these, 7,136 core non-toll projects and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The complete annexure to the Ministry of Road Transport and Highways' Rajya Sabha answer dated 1 April 2026 names 15 prioritised Multi-Modal Logistics Park locations. All 15 pass canonical identity, category, state, navigation, India-bounds, coordinate-order, administrative containment, and cross-catalog duplicate review.

The answer supports six source-dated Under implementation records, five Appraisal stage records, three DPR in progress records, and one DPR proposed record. These labels are not carried forward as present-day status. Four authority-estimated cargo-capacity quantities are not treated as current throughput, achieved capacity, or project cost.

The Ministry remains the exact Reporting ministry. The source does not establish an individual project owner, authority, executing agency, sponsor, contractor, concessionaire, or operator. Cost, expenditure, award value, payment, physical progress, completion, operation, audit findings, disputes, and full procurement histories remain explicitly missing.

The answer publishes no coordinate or geometry. Every displayed position is a separately attributed OpenStreetMap city or locality representative point. It is not the MMLP site, facility extent, land parcel, transport connection, surveyed coordinate, or project geometry.

Release 80 PNGRB pipeline inventory

Release 80 contained 7,210 accepted source-backed records. Of these, 7,121 core non-toll projects and infrastructure assets counted toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remained searchable and directly addressable through their explicit layer.

The reviewed source universe contains 65 stable pipeline identities across the Petroleum and Natural Gas Regulatory Board's April 2026 natural-gas and petroleum-product MIS reports. Fifty-six pass identity, category, state scope, representative navigation, bounds, coordinate-order, containment, and duplicate review. Nine source rows link to existing canonical PAIMANA records rather than creating a second project.

Thirty-three accepted records belong to the natural-gas family and 23 to the petroleum-product family. The March 2026 physical-state table supports 38 dated Operational, nine Partially commissioned, and nine Under construction observations. These states are not carried forward as present-day operation, availability, completion, or condition. April 2026 supply or throughput and utilization values remain separately typed monthly observations where PNGRB publishes them.

The source column is literally Entity. The product stores the exact role as PNGRB-reported entity and does not relabel it as owner, operator, authority, implementing agency, contractor, concessionaire, or sponsor. Project cost, expenditure, award value, payment, final cost, tender, contract, package, amendment, dispute, audit finding, and contractor remain explicitly missing from this baseline lane.

PNGRB publishes state scope and named pipeline identities but no reusable route geometry in the reviewed reports. Each displayed point is therefore a separately attributed OpenStreetMap endpoint or administrative navigation anchor. It is not the pipeline route, midpoint, station, right of way, surveyed site, or project extent.

D1 retains the complete 65-row disposition ledger, four source fingerprints, 56 baseline maturity rows, nine deduplication decisions, exact source links and roles, typed observations, and representative spatial records. Thirteen ordered discovery queries match the versioned static read model exactly, while static browse and evidence reads remain the independently healthy fallback.

Release 79 state-balanced AMRUT 2.0 inventory

The public catalog contains 7,154 accepted source-backed records. Of these, 7,065 core non-toll projects and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

A complete 18 July 2026 refresh of the Ministry of Housing and Urban Affairs AMRUT 2.0 public directory retrieved 3,376 ULB pages and 8,769 project rows without a failed page. Release 79 accepts a bounded state-balanced batch of 250 Water Supply and 250 Sewerage or Septage records across 31 states and union territories. The remaining candidate dispositions stay auditable and do not inflate the catalog.

The source's literal Project Cost quantity is not relabeled as original estimate, sanctioned cost, tender value, award, revised cost, expenditure, actual cost, or final cost. Its milestone is an undated portal label observed on the retrieval date, not an independently verified current progress, award date, completion date, or operating state.

The directory names a ULB but publishes no project coordinate or geometry. The 500 records therefore use 374 separately attributed cached OpenStreetMap locality anchors. Records from the same ULB deliberately share coordinates without jitter. Every marker remains a Locality representative point, not the project site, network, package extent, corridor, facility point, or surveyed geometry.

Twenty-one source tuples changed since the earlier audit and remain explicit revisions. Three unresolved ULB navigation gaps remain needs-review. D1 retains the complete 8,769-row ledger, the exact 500 accepted project identities, source fingerprints, separate source and navigation evidence, representative spatial records, and baseline maturity rows.

Release 78 CEA under-construction TBCB transmission projects

The public catalog contains 6,654 accepted source-backed records. Of these, 6,565 core non-toll projects and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The Central Electricity Authority's May 2026 Tariff Based Competitive Bidding under-construction report contains 86 top-level project rows and 301 nested component rows across 19 pages. Release 78 accepts 47 distinct project identities after complete-table extraction, catalog identity review, state-valid representative-place review, India bounds, and independent state and district containment. Thirty-nine rows resolve to existing PAIMANA canonical projects and remain auditable duplicate dispositions rather than inflating the catalog.

Under construction is a source-dated May 2026 report classification, not a claim carried forward beyond that reporting month. Component foundations, erection, stringing, civil works, equipment receipt, and remarks have different physical denominators and are not averaged into a synthetic whole-project progress percentage.

The report's Executing Agency role stays exact. Its Cost (Rs. Cr) quantity is not relabelled as original, sanctioned, awarded, revised, spent, paid, actual, or final cost. Original and anticipated target months remain separate typed dates. CEA states that the indicative report must not be used as a ground to claim time or cost overrun and does not relieve a transmission service provider of its agreement obligations.

CEA supplies no reusable coordinate or line geometry. The 47 accepted projects use 41 separately attributed OpenStreetMap representative navigation anchors from source-named places. Every marker remains a Corridor representative point, not a CEA coordinate, substation, route midpoint, tower, right-of-way, surveyed alignment, or complete project extent.

D1 retains the complete 86-row candidate ledger, 39 duplicate links and deduplication decisions, one fingerprinted official report, 47 accepted project and maturity rows, typed schedule and cost observations, source-dated lifecycle events, exact organization roles, source links, and spatial records.

Release 77 CEA under-construction RTM transmission projects

Release 77 contained 6,607 accepted source-backed records. Of these, 6,518 core non-toll projects and infrastructure assets counted toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remained searchable and directly addressable, but appeared in normal browsing only through their explicit layer.

The Central Electricity Authority's May 2026 Regulated Tariff Mechanism under-construction report contains 70 top-level transmission-project rows and 117 nested component rows. Release 77 accepts 58 unique project identities after complete-table extraction, catalog deduplication, state-valid representative-place review, India bounds, and independent state and district containment. Six rows resolve to existing PAIMANA projects and remain duplicate dispositions. Five source-named places lack a state-valid reusable navigation result, while one Dholera augmentation remains held because the parent or component relationship is not sufficiently established.

Under construction is a source-dated May 2026 report classification, not a claim carried forward beyond that reporting month. Component foundations, erection, stringing, civil works, equipment receipt, and remarks have different physical denominators and are not averaged into a synthetic whole-project progress percentage.

The report's Executing Agency role stays exact. Its Cost (Rs. Cr) quantity is not relabelled as original, sanctioned, awarded, revised, spent, paid, actual, or final cost. Original and anticipated target months remain separate typed dates, and their deterministic calendar-month difference is not presented as an overrun finding because CEA states that the indicative report must not be used as a ground to claim time or cost overrun.

CEA supplies no reusable coordinate or line geometry. The 58 accepted projects use 47 separately attributed OpenStreetMap representative navigation anchors from source-named places. Every marker remains a Corridor representative point, not a CEA coordinate, substation, route midpoint, tower, right-of-way, surveyed alignment, or complete project extent.

D1 retains the complete 70-row candidate ledger, six duplicate links, six needs-review rows, one fingerprinted official report, 58 accepted project and maturity rows, typed schedule and cost observations, source-dated lifecycle events, exact organization roles, source links, and spatial records.

Release 75 freestanding infrastructure pictograms

The 16-category taxonomy, every category assignment, source coordinate, precision class, evidence record, discovery filter, lifecycle state, result, URL, dossier, toll default, and D1 inventory row remain unchanged. Release 75 replaces only the category-icon presentation.

Each category now uses a freestanding professional Lucide pictogram with a stable reviewed color, silhouette-following warm-white contour, and restrained neutral shadow. There is no permanent enclosing category disc, square, diamond, or capsule. Major records use a larger icon footprint than Facilities, while the invisible pointer and touch target remains 28 to 32 pixels.

Spatial precision remains independent and subordinate. Exact points use one short attached line, official facilities use a double line, representative localities use a dotted line, corridors use a short tailed line, mine representatives use a dotted anchored treatment, and unresolved locations add a question mark. These marks do not change a coordinate or imply a surveyed site. The map key, accessible results, preview, Project Brief, and evidence record remain authoritative.

Selection and keyboard focus use a temporary silhouette-following neutral outline around a larger version of the same category pictogram. Metric lenses may add a separate semantic ring. Neither treatment recolors the category, erases precision, or turns an attention state into a verdict.

Subtype presentation is selected only from reviewed source-type or sector fields. The current catalog can distinguish source-backed submarine cable and telecom records inside Digital & Telecom. Irrigation, renewable-generation, and data-centre assets are prepared for future evidence but remain dormant until a reviewed record explicitly supports the subtype. Subtype presentation never creates a new filter category.

The map continues to use native MapLibre symbol layers and feature state. No project receives a DOM marker, generated displacement, cluster, count marker, or coordinate jitter. A small source-locked category-colored point remains the honest first-paint and icon-failure fallback while the icon sprite settles; search, results, dossiers, and evidence remain usable without map icons or the basemap.

Release 76 CEA completed-TBCB transmission projects

The public catalog contains 6,549 accepted source-backed records. Of these, 6,460 core non-toll projects and infrastructure assets count toward the 10,000-record objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The Central Electricity Authority's May 2026 completed Tariff Based Competitive Bidding report contains 79 top-level transmission-project rows. Release 76 accepts 78 unique project identities after complete-table extraction, normalized-name catalog review, representative-place review, India bounds, state context, and independent state and district containment. Nangalbibra or Nongalbibra remains needs review because no state-valid navigation result was located; no substitute centroid or invented route closes the gap.

The report's Executing Agency role stays exact. It is not silently relabelled as owner, sponsoring ministry, EPC contractor, concessionaire, operator, or project company. The report's Cost (Rs. Cr) quantity is separately typed and is not called original, sanctioned, awarded, revised, spent, paid, actual, or final cost. No cost-change calculation is supported from that single quantity.

The product shows the original scheduled COD, actual COD, and their deterministic calendar-month difference. CEA states that this indicative report must not be used as a ground to claim time or cost overrun, so the interface does not convert the difference into an overrun allegation. A Completed report section and Actual COD do not establish present operation, availability, asset condition, utilization, ownership, or benefit realization.

CEA supplies project names, state columns, dates, agency, cost, and component scope but no reusable coordinate or line geometry. The 78 accepted records use 57 separately attributed OpenStreetMap representative navigation anchors from source-named places. Every marker remains a Corridor representative point, not a CEA coordinate, substation, route midpoint, tower, right-of-way, surveyed alignment, or complete project extent.

D1 retains the complete 79-row candidate ledger, one held disposition, source fingerprint, 78 accepted project and maturity rows, typed schedule and cost observations, lifecycle events, exact organization roles, source links, and spatial records. Static browse and evidence reads remain the independently healthy public fallback while production D1 reads stay behind parity and failure-isolation checks.

Release 74 Amrit Bharat Station Scheme inventory

The public catalog contains 6,471 accepted source-backed records. Of these, 6,382 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The Ministry of Railways list published through PIB provides a complete 1,309-position Amrit Bharat Station Scheme source universe. Release 74 accepts 1,095 station-development records after exact official identity parsing, conservative normalized-name matching to one railway-station navigation item, India bounds, independent state containment, and cross-catalog review. Seventy-one candidates remain needs review and 143 remain pending. Three source rows publish one fewer safely delimited station name than their state count; no identity is invented to close those gaps.

The scheme list establishes that a named station was identified for development as of 8 December 2023. A later official list supports a completed lifecycle observation for 141 accepted station records as of 12 March 2026. Identification is not construction start, award, completion, commissioning, inauguration, opening, or present operation. Completion does not establish final cost, contractor, present operation, or the scope of every work. The 954 records without a matched completion observation carry an explicit newer-evidence warning rather than an inferred current status.

Each accepted station uses a separately attributed Wikidata, GeoNames, or OpenStreetMap representative navigation point after bounded identity matching and state validation. All 1,224 accepted Amrit Bharat station points match their source state. That point is not an official Ministry of Railways coordinate, station boundary, platform, station building, parcel, construction worksite, or project footprint. It does not upgrade source precision or prove that every scheme work occurs at that point.

All station developments are Infrastructure facilities. They appear in the default non-toll All-records view and the Facilities layer, remain searchable and filterable by Railway and lifecycle, and do not enter the Major-project layer. Five possible overlaps with earlier station or terminal development projects remain review-held rather than merged or double-counted.

D1 retains the complete candidate ledger, source fingerprints, lifecycle crosswalk, aliases, exact Scheme-launching ministry role, source links, evidence observations, spatial records, validation boundaries, and maturity rows. The public static discovery and detail paths remain an independently healthy fallback while production D1 reads continue to be parity-tested incrementally.

Selected-record data is partitioned into 256 deterministic static shards and loaded only after selection. The complete 51,631,857-byte detail catalog no longer executes inside the edge worker at startup. This reduced the exact minified worker from a failed 50,513,456-byte deployment candidate to 734,512 bytes without removing a record, source, spatial limit, Project Brief, specialist section, or map-independent fallback.

Release 73 MoPSW National Waterway systems

The public catalog contains 5,376 accepted source-backed records. Of these, 5,287 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

The Ministry of Ports, Shipping and Waterways list supplies 111 stable National Waterway identities, source component descriptions, published component lengths, and state or multi-state scope. The National Waterways Act, 2016 supplies the statutory boundary. National Waterways 1 to 5 retain their earlier declarations; the Act declares the scheduled waterways numbered 6 to 111.

Legal designation is not an operational claim. A listed or declared National Waterway does not establish current operationalization, navigability, cargo or passenger traffic, dredging, terminal access, channel condition, construction, progress, completion, commissioning, opening, operator, contractor, cost, schedule, or present condition.

The reviewed Ministry material does not expose a reusable coordinate or route dataset. A one-time cached OpenStreetMap Nominatim review supplies separately attributed navigation geography under ODbL. Sixty-eight candidates have a named waterway feature that matches the reviewed hydronym or alternate name, lies in a source-named state, and passes India-bounds and coordinate-order checks. Forty-three candidates remain needs review. No city centroid, district point, generated line, synthetic route, or invented coordinate replaces the gap.

Every published position remains a Corridor representative point on one source-named component. It is not an official Ministry coordinate, waterway centre, endpoint, surveyed site, channel limit, navigable extent, complete route, or proof that the full source-published length lies in the mapped state. Source component lengths remain separately typed and are not converted into browser-derived route geometry.

National Waterway systems remain separate from development projects on those systems. National Waterway 1 is linked to the existing Jal Marg Vikas Project without merging the entities or transferring lifecycle, money, schedule, organization, progress, or geometry claims.

D1 retains all 111 candidate dispositions, two fingerprinted official sources, 68 baseline records, separately attributed navigation sources, 204 project-source links, 136 typed observations, 68 exact source-publisher roles, aliases, spatial evidence, and the reviewed NW-1 relationship. Static discovery and evidence reads remain the independent public fallback while production D1 reads stay behind parity and failure-isolation gates.

Release 72 MoPSW major-port assets

Release 72 contained 5,308 accepted source-backed records. Of these, 5,219 core non-toll projects and infrastructure assets counted toward the 10,000-project objective. The unchanged 89 reviewed toll plazas remained searchable and directly addressable, but appeared in normal browsing only through their explicit layer.

The source universe is the 14 major ports designated by the Ministry of Ports, Shipping and Waterways. Twelve operational port assets have both a published topography row and positive May 2026 cargo in the reviewed advance estimate, so they enter as baseline inventory records. Vadhavan and Galathea Bay remain needs-review construction candidates because the reviewed tables do not establish an operational location and cargo row for them.

The lifecycle label is deliberately narrow. Positive May 2026 cargo supports an operational observation dated 31 May 2026. It does not establish uninterrupted current operation, present service quality, capacity, utilization, construction completion, commissioning, ownership, operator, concessionaire, financial performance, or condition after that date.

The Basic Port Statistics table publishes degree-minute-second or degree-minute latitude and longitude but does not state a geodetic datum. The product preserves the original text, normalizes it into longitude-latitude order for WGS 84 display, and does not upgrade the result into a surveyed claim. Nine rows are official port reference points. Deendayal, Mumbai, and Syama Prasad Mookerjee use representative anchors because the source describes multiple terminals, docks, or location pairs.

Five new points fall within the reviewed state and district land polygons. Seven official coastal points sit outside those land-only polygons and therefore retain an explicit coastal exception with no containing district. The map does not move them inland or invent a district. No point is a port limit, berth or terminal boundary, cadastral parcel, surveyed centre, construction site, or complete asset extent.

Port assets and development projects remain separate entities. Seven earlier port projects stay intact, and five reviewed relationships connect Deendayal, New Mangalore, Paradip, and Syama Prasad Mookerjee assets to source-listed terminal, berth, jetty, or pipeline works without merging their identities or evidence.

D1 retains the full 14-row candidate ledger, two immutable Ministry documents, exact candidate dispositions, 12 baseline maturity rows, 24 project-source links, 24 field observations, official aliases, exact major-port authority or company roles, and the reviewed relationships. Static browse and evidence reads remain the independent public fallback while production D1 reads stay staged.

Release 71 DGCA public-use aerodrome assets

The public catalog contains 5,296 accepted source-backed records. Of these, 5,207 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

This release adds 105 airport assets from the Directorate General of Civil Aviation's public-use aerodrome register updated on 17 July 2026. The lifecycle label means only that the reviewed DGCA row reports a public-use licence valid on that date. It does not establish current flights, scheduled passenger service, operation, ownership, commissioning, construction completion, capacity, or conditions after the validity evidence.

The 110-row candidate ledger remains complete: 105 accepted, four needing review, and one older Thiruvananthapuram observation linked as a duplicate. Aizawl and Rewa have passed validity dates, Bhogapuram lacks a licence number and validity dates, and Shravasti has a current DGCA and AAI licence-identity conflict. These rows remain outside the accepted count.

DGCA publishes no coordinate or geometry in the reviewed register. One hundred three accepted points therefore come from a pinned public-domain OurAirports snapshot, while Datia and Pantnagar use cached OpenStreetMap Nominatim results under ODbL. Each remains a representative facility point, not a DGCA coordinate, official aerodrome reference point, runway point, terminal coordinate, parcel, construction site, or airport polygon.

All 5,296 map positions pass the same state and district checks. The earlier 28 source-state differences remain explicit, and Kakari Mahudi Dam remains the one position returned by two districts. All 105 new airport points match their declared state context and each returns one containing district. Containment validates the displayed point, not an airport's complete extent or served area.

Existing airport development projects and the new airport assets remain separate entity kinds. Fifteen confident relationship candidates are retained for review, but no licence becomes a project completion state and no similar name merges an asset with a runway, terminal, airport-development package, or construction project.

D1 now retains five auditable source control planes and imports the 105 accepted assets through an incremental idempotent migration. The same twelve ordered discovery queries match the versioned static read models exactly. Public browsing and evidence access still retain the static fallback while production database reads remain staged.

Release 70 historical CEA generation assets

The public catalog contains 5,191 accepted source-backed records. Of these, 5,102 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The unchanged 89 reviewed toll plazas remain searchable and directly addressable, but appear in normal browsing only through their explicit layer.

This release adds 102 non-hydro station assets from the Central Electricity Authority's January 2024 monthly generation table published through the Open Government Data Platform. All accepted rows have a positive January 2024 generation observation. That dated observation is not carried forward as proof of present operation, availability, ownership, commissioning, or performance.

The 526-row CEA candidate ledger remains complete: 102 accepted, 46 needing review, 98 hydro rows deferred until they can be reconciled with existing NWIC powerhouse assets, and 280 pending because no conservative navigation match was available. Candidate, held, deferred, and pending rows do not count toward the catalog.

The CEA source publishes station name and state but no coordinate. Each accepted map point therefore comes from a separately attributed frozen Wikidata CC0 navigation match that passed a 0.90 name score, a 0.08 next-candidate margin, unique item use, India bounds, and independent state containment. The point remains representative and is not an official CEA coordinate, surveyed site, plant boundary, unit location, or exact project footprint.

All 5,191 map positions pass the same state and district checks. The earlier 28 source-state differences remain explicit, and Kakari Mahudi Dam remains the one position returned by two districts. The 102 new CEA records add no new state conflict or ambiguous district result.

D1 now retains four auditable source control planes and imports the 102 accepted assets through an incremental idempotent migration. The same twelve ordered discovery queries match the versioned static read models exactly. Public browsing and evidence access still retain the static fallback while production database reads remain staged.

Release 69 five-thousand inventory checkpoint

The public catalog contains 5,089 accepted source-backed records. Exactly 5,000 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The same 89 reviewed toll plazas remain searchable, directly addressable, and available through their explicit layer, but are reported separately.

The checkpoint adds 2,500 records through three independently reviewable contributions: 1,873 more National Water Informatics Centre water assets, 309 PAIMANA road or railway projects, and 318 AMRUT 2.0 water-supply or sewerage projects. Their source identity, record kind, category, evidence date, spatial precision, validation result, and missing fields remain distinct.

D1 retains the complete 7,024-row NWIC, 1,987-row PAIMANA, and 8,769-row AMRUT candidate ledgers. Only accepted canonical records count. Three NWIC duplicates, 184 NWIC review holds, 3,375 pending NWIC rows, 10 PAIMANA duplicates, 722 PAIMANA review holds, 94 rejected or deferred PAIMANA rows, 18 AMRUT anchor collisions, and 8,398 pending AMRUT rows remain outside the accepted count.

AMRUT geography uses previously reviewed source-backed urban-local-body anchors. A shared locality anchor does not become an exact work site, treatment-plant footprint, pipeline, drain, or sewer alignment. PAIMANA candidates without a state-consistent spatial lookup remain held. NWIC coordinates retain their published CRS, deterministic WGS 84 transformation, and 28 September 2022 freshness boundary.

All 5,089 positions pass independent state and district checks. Twenty-eight source-state differences remain explicit. One boundary position, Kakari Mahudi Dam, is returned by both Arvalli and Mahisagar districts, so the record retains a multi-district validation state and no chosen district.

The checkpoint has measurable coverage in all 16 maintained categories and every state or union territory, but it is not balanced national completeness. Water Resources contains 3,202 records, while Digital and Telecom contains 3, Logistics and Inland Transport 4, Ports 7, and Industrial Corridors 8. Category and source-date gaps remain visible rather than being disguised by totals.

The healthy discovery path returns a 24-row first page plus exact active IDs instead of full dossiers. The complete current-fact artifact remains a failure fallback; selected briefs and specialist evidence load independently. Release validation also removed a shared runtime import that embedded the full source catalog in the client bundle, reducing the compiled InfrastructureMap chunk from 41.3 MB to about 227 KB. D1 is the canonical structured inventory and control plane, but public database reads remain staged until production parity, latency, caching, rollback, and failure-isolation evidence is complete.

Release 68 second-source inventory checkpoint

The public catalog contains 2,589 accepted source-backed records. Exactly 2,500 core non-toll projects and infrastructure assets count toward the 10,000-project objective. The same 89 reviewed toll plazas remain searchable, directly addressable, and available through their explicit layer, but are reported separately.

Release 68 adds 1,589 baseline inventory assets from three National Water Informatics Centre spatial resources published through India's Open Government Data platform: 221 dams, 282 powerhouses, and 1,086 barrages, weirs, or anicuts. Their stable source identifiers, archive fingerprints, projected coordinates, published CRS, deterministic WGS 84 transformations, source attributes, and limitations remain auditable.

The 7,024-row NWIC ledger records 1,589 accepted or published candidates, 3 reviewed duplicates, 184 needs-review rows, and 5,248 pending rows. Coordinate disagreements, JK-coded geography requiring current J&K or Ladakh review, missing state scope, and two border-sensitive positions remain outside the accepted count. The unchanged PAIMANA ledger remains a separate source control plane.

The NWIC resources were updated on 28 September 2022. They establish source-listed assets and official reference points as of that source date. They do not establish present operation, current ownership, implementing authority, contractor, cost, progress, exact facility footprint, or current lifecycle. Every published NWIC asset therefore carries a newer-evidence warning rather than an inferred current status.

All 2,589 map positions pass state and district containment. Twenty-five source-state differences remain explicit. Containment is a validation signal for the displayed point, not surveyed accuracy or proof of a complete corridor, reservoir, facility, or project extent.

D1 retains both source registries, candidate ledgers, maturity, aliases, source links, observations, spatial transformations, and deduplication decisions. The public discovery and detail paths continue to use versioned static read models as a tested fallback while production database reads remain staged behind parity, latency, caching, rollback, and failure-isolation checks.

When more than 1,200 records match, the national view deterministically prioritizes 700 individual markers using a three-to-one established-project and baseline-asset mix plus category and state balance. The exact result count remains visible, and zooming or filtering progressively reveals every matching source coordinate. This is not clustering, random sampling, coordinate movement, or a claim that hidden-at-this-zoom records are absent.

Release 67 inventory checkpoint and D1 candidate ledger

The public catalog contains 1,000 accepted source-backed records. The long-term 10,000-project count is 911 core non-toll records; 89 toll plazas remain searchable, directly addressable, and available through their explicit layer but are reported separately rather than used to inflate the target.

Release 67 adds 382 PAIMANA records through separately reviewable road, railway, and cross-sector manifests. The cumulative PAIMANA lane now contains 852 accepted records. All additions preserve official project code and source fields, evidence-dated lifecycle, typed money and schedule fields, representative geography limitations, and explicit missing evidence.

The D1 control plane retains all 1,987 PAIMANA source rows: 852 accepted or published, 10 duplicate, 136 needs review, 94 rejected or deferred, and 895 pending. Candidate, duplicate, rejected, deferred, needs-review, alias, package, and phase rows do not count as canonical projects unless they independently satisfy the baseline contract.

Inventory maturity is explicit. A baseline record has validated identity, category, record kind, geography and precision, lifecycle as of an evidence date, and an authoritative source. Validated-detail and deep-evidence tiers add compatible schedule, money, progress, role, contract, payment, audit, milestone, or geometry evidence where published. An absent deep field remains missing.

The static public discovery and detail read models remain the independent fallback. D1 now exposes the checkpoint ledger, maturity counts, deduplication decisions, and twelve ordered-query parity checks. A full public D1 discovery cutover remains staged until production query behavior, latency, caching, rollback, and failure isolation are proven.

Release 66 breadth and twelve-query D1 parity boundary

The public catalog now contains 618 source-backed records: 494 Major projects or mine assets and 124 optional Facilities or local schemes. The default non-toll scope contains 529 records, while the same 89 reviewed toll plazas remain available only through the explicit Toll plazas layer. One hundred additional Major records extend the cumulative PAIMANA lane from 370 to 470 without changing any earlier identity, coordinate, precision label, source field, or evidence trail.

The batch comes from the already reviewed 1,987-row May 2026 PAIMANA universe. It adds 25 railway systems, 20 national-highway or road projects, 30 public healthcare or higher-education projects, 15 refinery or fuel-system projects, and 10 steel or mineral-processing development projects across 26 state tokens. Classification is based on the official sector and source-published physical scope, not cost or name alone. Steel and mineral-processing development projects remain separate from Mine assets and do not establish present mine or plant operation.

Each new point is a cached, reviewed navigation representation with source query, retrieval date, original coordinates, WGS 84, spatial type, precision, and limitations. Facility points, representative localities, and corridor anchors remain different meanings. Twenty-four new and 55 cumulative genuine shared-anchor cases retain identical reviewed coordinates and explicit collision records rather than jitter, displacement, or invented geometry.

The independent state and district checks now cover all 618 positions. Six hundred twelve agree with their declared source-state scope, while the same six earlier NHAI conflicts remain visible. Every position resolves to one district across 344 distinct Local Government Directory district-code pairs.

A Sites D1 binding, migration ledger, and idempotent 618-record import stage the same project, current-fact, spatial, state-facet, and awareness rows for semantic-parity review. Local validation and the public catalog-status endpoint compare exact ordered project IDs and SHA-256 digests for twelve representative queries, including total and default scope, category by state, inauguration, evidence tier, compatible cost movement, environmental utilities, text search, and the expanded Railway and Urban Infrastructure categories. The public map and dossier remain independently available from versioned static read models; D1 is not yet the authoritative public query path, and R2 remains unbound.

Release 62 Polavaram audit, contract, and funding boundary

Release 62 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens Polavaram Irrigation Project with the Comptroller and Auditor General of India's Performance Audit Report No. 4 of 2025 and two Ministry of Jal Shakti parliamentary replies. The record moves from Timeline validated to Contract-linked without moving its representative Polavaram locality point.

The CAG audit was tabled on 6 March 2026. Its performance scope covers 2017-18 to 2022-23, its project-status comparisons are as of March 2023, and the reviewed government replies extend through November 2024. The audit is therefore a historical accountability source, not evidence of present construction state. PAIMANA remains the current record source for 86.14% reported progress, the ₹10,151.04 crore Original Cost, ₹55,548.9 crore Revised Cost, ₹27,061.89 crore Cumulative Expenditure, and March 2026 Revised DoC.

Six audit observations retain the exact CAG chapter and page range, what the audit reported, why it was highlighted, any amount and its definition, the government reply, the CAG response, the recommendation, the audit cutoff, and an interpretation boundary. The interface does not convert those observations into a product-authored claim of corruption, misconduct, blame, present failure, or final loss.

The money records remain semantically separate. The CAG records ₹47,725.74 crore as the Revised Cost Committee recommendation at 2017-18 price level, which had not been accepted by the audit's March 2023 cutoff. The Ministry's December 2025 answer separately publishes rounded figures of ₹30,436 crore for the approved revised cost estimate at 2017-18 price level, ₹12,157 crore for additional central assistance, and ₹20,658 crore for cumulative central assistance. None is silently substituted for PAIMANA's differently defined current record.

The contract history retains TRANSSTROY-JSC-EC-UES JV for the March 2013 original Main Dam contract, Navayuga for the February 2018 agency change and separately described powerhouse electromechanical role, and Megha Engineering & Infrastructures Limited for the November 2019 ₹4,359.11 crore balance Main Dam and powerhouse award. The rows do not establish a complete current contractor register, certified payment, final account, current performance finding, or final cost.

The official CAG landing page and PDFs were reviewed through the public document viewer. Stable local file retrieval from the CAG host was unavailable during this release, so the ledger records page-level review and the access limitation but does not claim a local SHA-256 fingerprint. No reviewed source publishes an authoritative whole-project point, component geometry, reservoir or command-area polygon, or current rehabilitation outcome. The unchanged map point remains representative locality context.

Release 61 Bikaner IV Part A procurement, agreement, tariff, licence, and monitoring boundary

Release 61 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens Bikaner IV renewable-energy transmission Part A with six fingerprinted official PDFs from RECPDCL, CERC, and CEA. The record moves from Timeline validated to Contract-linked without moving its representative Bikaner corridor point.

The evidence retains four distinct stages: RECPDCL's 6 June 2024 RFP, its 17 September 2024 Amendment VI, the final ranked bid result, and CERC's records of the Letter of Intent, agreement and SPV-transfer events, tariff adoption, and final transmission licence. Amendment VI scheduled the Letter of Intent for 16 October and SPV transfer for 28 October; CERC records the events on 21 October and 11 November. These dated differences do not establish breach, fault, or construction delay.

Money meanings remain separate. CERC publishes ₹4,461.05 crore as the bidding-process estimated project cost. CEA and PAIMANA publish ₹5,969 crore as scheme cost. CERC and RECPDCL record ₹587.906 crore per year as POWERGRID's selected and adopted annual transmission charges, while CERC records ₹657.179 crore per year as estimated annual transmission charges and ₹157.5 crore as the Contract Performance Guarantee. These quantities and PAIMANA cumulative expenditure are not a compatible cost-revision series and are not relabeled as EPC value, payment, actual cost, or final cost.

CERC's 28 April 2025 order grants POWERGRID Bikaner IV Transmission Limited a 25-year final inter-State transmission licence. It does not name an EPC contractor or award and does not prove construction completion, energisation, commissioning, present operation, tariff recovery, or final outturn. CERC's separately condoned nine-day portal-fee filing delay is a regulatory filing event, not a finding of project delay.

CEA's May 2026 row names six components: the 12,000 MVA Bikaner-IV pooling station, four transmission-line elements, and the Bikaner-IV STATCOM. Their circuit-kilometre, location, foundation, erection, civil-work, equipment, land, right-of-way, and forest-clearance fields remain literal component evidence. They are not averaged into a synthetic project percentage. November 2026 and April 2027 remain source-published target months, not an inferred contractual breach, fault, or final completion state.

Part A and the separately published Siwani Part B record share source-listed interfaces but remain distinct project entities, SPVs, costs, tariffs, licences, and monitoring rows. No reviewed source publishes a complete line route, right-of-way geometry, pooling-station coordinate, equipment footprint, or project extent. The existing Bikaner point remains an unchanged representative corridor navigation anchor. A newer authoritative monitoring source is required for conditions after May 2026.

Release 60 Siwani procurement, agreement, tariff, licence, and monitoring boundary

Release 60 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens Bikaner REZ Phase IV Part B with five fingerprinted official PDFs from RECPDCL, CERC, and CEA. The record moves from Timeline validated to Contract-linked without moving its representative Siwani corridor point.

The evidence retains four distinct stages: RECPDCL's 6 June 2024 RFP, its 2 September 2024 Amendment IV, CERC's record of the POWERGRID bid result and 21 October Letter of Intent, and the later agreement, SPV-transfer, tariff-adoption, and final transmission-licence events. The signed Share Purchase Agreement, Transmission Service Agreement, SPV acquisition price, and EPC tender or award remain unlinked.

Money meanings remain separate. CEA and PAIMANA publish ₹5,357 crore as scheme cost. CERC records ₹509.313 crore per year as POWERGRID's selected and adopted annual transmission charges, ₹564.734 crore per year as estimated annual transmission charges for bid evaluation, and ₹139.5 crore as the Contract Performance Guarantee. These quantities and PAIMANA cumulative expenditure are not a compatible cost-revision series and are not relabeled as EPC value, payment, actual cost, or final cost.

CERC's 26 March 2025 order grants POWERGRID Siwani Transmission Limited a 25-year final inter-State transmission licence and directs competitive bidding for the EPC contract. It does not name an EPC contractor or award and does not prove construction completion, energisation, commissioning, present operation, tariff recovery, or final outturn.

CEA's May 2026 row names five components: the 9,000 MVA Siwani substation, three transmission lines, and the Siwani STATCOM. Their circuit-kilometre, location, foundation, erection, civil-work, equipment, land, right-of-way, and forest-clearance fields remain literal component evidence. They are not averaged into a synthetic project percentage. November 2026 and June 2027 remain source-published target months, not an inferred contractual breach, fault, or final completion state.

No reviewed source publishes a complete line route, right-of-way geometry, substation-site coordinate, equipment footprint, or project extent. The existing Siwani point remains an unchanged representative corridor navigation anchor. A newer authoritative monitoring source is required for conditions after May 2026.

Release 59 Dhubri-Phulbari bridge approval, procurement, award, and financing boundary

Release 59 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens the existing Dhubri-Phulbari Brahmaputra Bridge with seven official sources from PIB, NHIDCL, and JICA. The record moves from Timeline validated to Contract-linked and adds Meghalaya to its official discovery scope without moving its representative Dhubri corridor point.

The 28 February 2019 Cabinet release publishes a 19.282 km four-lane bridge and approaches, ₹3,548 crore as civil-construction cost, ₹4,997.04 crore as total capital cost, and JICA loan assistance. NHIDCL's fingerprinted status register separately publishes ₹3,165.99 crore as Awarded Cost, names Larsen & Toubro Limited, Heavy Civil Infrastructure IC as contractor, and records the 24 November 2020 award and 15 December 2020 appointed date.

Those quantities remain separate from PAIMANA's ₹2,837.17 crore Cumulative Expenditure and JICA's JPY 34,537 million second-tranche loan agreement, for which JICA provides an approximate ₹1,946 crore equivalent. Total Project Cost, civil-construction cost, Awarded Cost, cumulative expenditure, financial-progress percentage, financing, payment, actual cost, and final cost are not interchangeable.

The procurement trail keeps each stage literal. NHIDCL annulled the first bidding process on 10 January 2020 after one responsive bid and said it would re-invite the tender. Its 3 November 2020 technical evaluation lists Larsen & Toubro and S.P. Singla Constructions as technically responsive. Technical responsiveness is not an award, and annulment does not establish fault, misconduct, corruption, or project delay attributable to a party.

PAIMANA reports 76.01% physical progress with a 24 June 2026 reporting cutoff. NHIDCL's current HTML page separately publishes 76% physical and 74% financial progress while retaining 12 September 2028 as original and likely completion, but it has no row-level status date. The NHIDCL values therefore remain retrieval-dated and do not become a later measurement, payment amount, regression, or completion finding.

The official project evidence spans Assam and Meghalaya. The map still uses the previously reviewed Dhubri OpenStreetMap anchor because no authoritative bridge alignment, approach geometry, midpoint, surveyed site, right of way, or project extent was located. The signed contract, amendments, payment certificates, final account, certified completion, opening, present operation, disputes, audits, final cost, and authoritative corruption findings remain missing.

Release 58 Dholera airport approval, package-award, and status boundary

Release 58 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens Dholera International Greenfield Airport with a Cabinet approval, a fingerprinted Ministry of Civil Aviation annual report, three Airports Authority of India contract-award rows, and a 14 July 2026 PIB review. The record moves from Timeline validated to Contract-linked; no marker or geometry changes.

The 14 June 2022 Cabinet approval publishes ₹1,305 crore as the approved Phase I estimate and names Dholera International Airport Company Limited as implementing company. The MoCA Annual Report 2025-26 publishes ₹1,551 crore as Approved Cost; PAIMANA labels the same later quantity Revised Cost. The product shows the compatible ₹246 crore, 18.85% movement while retaining the exact labels. Neither quantity is actual or final cost.

AAI publishes three package awards: ₹636.4454249 crore to Varaha Infra Limited for operational-area works, ₹333 crore to Yashnand Engineers and Contractor Private Limited for the terminal complex, and ₹18.9880385 crore to Saumya Construction for the ASR-MSSR radar building. These are package contract values. They are not added into a new airport total, relabeled as payment or expenditure, or treated as actual or final cost.

Package roles remain exact. There is no invented whole-project contractor. The Cabinet approval also preserves AAI, Government of Gujarat, and NICDIT as 51%, 33%, and 16% shareholders in DIACL. Past source-scheduled package dates remain dated contract-award fields and do not by themselves establish actual completion, breach, a cause of movement, or contractor fault.

PAIMANA reports 84% physical progress in its May 2026 snapshot. PIB reports approximately 80% overall construction and 75% terminal-building progress on 14 July 2026, together with source-reported completion of the ancillary structures, ATC tower, primary runway, and taxiway. Both progress statements remain attributable; no regression is inferred because their scopes and rounding may differ.

PIB's September on-site-work direction and September or October 2026 operationalization aim, PAIMANA's September 2026 Revised DoC, and MoCA's FY 2026-27 expectation remain targets. They do not establish full construction completion, DGCA licensing, inauguration, opening to traffic, present operation, achieved throughput, or final outturn. The unchanged point remains a representative OpenStreetMap facility context, not an official airport coordinate, package polygon, runway geometry, or surveyed site.

Release 57 Bidar procurement, amendment, tariff, and monitoring boundary

Release 57 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens the existing Bidar 2,500 MW solar-zone transmission record with eight fingerprinted official sources across REC, CERC, and CEA. The record moves from Timeline validated to Contract-linked; no new marker or geometry is added.

The evidence establishes two distinct procurement cycles. REC's archive lists the initial RFP and Amendments I through XXI, then marks that process annulled. The 2023 RFP explicitly says the 2020 process was annulled and re-initiated. The second cycle reached Amendment XVII, three qualified bidders, POWERGRID's selected ₹2,368.09 million-per-year offer, Letter of Intent, CERC-recorded guarantee, Share Purchase Agreement and SPV acquisition events, and tariff adoption subject to the transmission licence.

An official chronology difference remains unresolved. CERC says the initial bidding process was annulled in September 2020 due to non-availability of LTA applications, while REC publishes Amendment XXI dated 23 August 2021. The product keeps both statements attributable and does not invent an exact annulment date or explanation.

The money types remain separate. CEA and PAIMANA publish ₹1,368 crore as scheme cost. Amendment XVII changes the bid bond from ₹29.80 crore to ₹34.36 crore and the Contract Performance Guarantee from ₹69.50 crore to ₹85.90 crore. CERC adopts ₹236.809 crore per year as the annual transmission charge and compares it with an estimated ₹355.893 crore per year. Security, qualification threshold, annual tariff, scheme cost, expenditure, payment, actual cost, and final cost are not one revision series.

CEA's May 2026 row adds the 479 circuit-kilometre Bidar–Maheshwaram line and 7,000 MVA pooling-station component with literal construction indicators, land, and right-of-way remarks. They remain component facts and are not averaged into a synthetic whole-project percentage or represented as invented line or site geometry. February 2026 and August 2026 stay source-published target months, not a contractual blame finding or proof of current status.

Release 56 Bhadla-Bikaner procurement, tariff, and monitoring boundary

Release 56 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens the existing Bhadla-III and Bikaner-III interconnection-strengthening record with four fingerprinted official PDFs: PFCCL's 5 July 2024 qualification notice and 30 August 2024 final result, CERC Order in Petition No. 342/AT/2024 dated 14 January 2025, and CEA's May 2026 under-construction transmission report. The record moves from Timeline validated to Contract-linked; no new marker or geometry is added.

PFCCL names four qualified bidders and publishes POWERGRID as L1 at ₹979.70 million per year. CERC records the full bid chronology, the same ₹979.70 million-per-year bid table and Bid Evaluation Committee recommendation, the 1 August 2024 Letter of Intent, the 30 August 2024 guarantee, Share Purchase Agreement, and SPV-acquisition events, and tariff adoption subject to the transmission licence.

The annual charge remains an explicit source conflict. PFCCL's final result, CERC's bid tables, and the committee recommendation publish ₹979.70 million per year. CERC paragraphs 27 and 28 instead call ₹979.90 million per year the lowest and adopted charge. Release 56 retains both quantities, their pages, and their evidentiary roles. It does not silently correct, average, or turn the ₹0.20 million-per-year difference into a project-cost movement.

Money definitions remain separate. CERC's factual matrix publishes ₹901.41 crore as Estimated Cost of the Project. CEA and PAIMANA publish ₹1,382 crore for the scheme. CERC also records a ₹30 crore Contract Performance Guarantee. The estimate, scheme cost, recurring annual charges, guarantee, acquisition consideration, payment, expenditure, actual cost, and final cost are distinct quantities; the sources do not establish a compatible original-versus-revised project-cost series.

CEA names one monitored component, the 254 circuit-kilometre Bhadla-III to Bikaner-III 765 kV double-circuit line, with 327 total locations, 232 foundations, and 100 erections. These are literal component fields, not a synthetic percentage or whole-project completion claim. CERC's scheduled 30 August 2026 COD, CEA's August 2026 original and March 2027 anticipated targets, and PAIMANA's March 2027 Revised DoC remain source-labeled schedule fields rather than an inferred contract amendment or delay finding.

The CERC order is a regulatory attestation, not the signed Share Purchase Agreement or Transmission Service Agreement. The complete RFP, signed instruments, acquisition price, amendments, final licence order, EPC packages, payments, disputes, audits, tariff recovery, actual cost, final cost, and status after May 2026 remain unlinked. The unchanged Bhadla village point remains a representative corridor anchor, not a pooling station, alignment, midpoint, right of way, or project extent.

Release 55 South Olpad procurement, tariff, and monitoring boundary

Release 55 preserves the 393-record catalog, 304-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every precision label. It deepens the existing Khavda Phase-IV 7 GW Part B record with three fingerprinted official PDFs: the PFCCL RFP published on 19 October 2023, CERC Order in Petition No. 493/AT/2024 dated 24 May 2025, and CEA's May 2026 under-construction transmission report. South Olpad moves from Timeline validated to Contract-linked; no new marker or geometry is added.

The CERC order records the bid chronology, three e-reverse-auction participants, POWERGRID as successful bidder, the 19 September 2024 Letter of Intent, the 15 October 2024 agreement and SPV-transfer events, a ₹100 crore Contract Performance Guarantee in favour of CTUIL, and adoption of ₹556.686 crore per year in annual transmission charges. The annual charge is a recurring tariff, and the guarantee is a contractual safeguard. Neither is project cost, EPC award value, expenditure, payment for completed work, actual cost, or final cost.

Money definitions remain separate. CERC's bidding matrix publishes ₹3,605.57 crore as an estimated project cost. CEA and PAIMANA separately publish ₹4,766 crore for the scheme. The sources do not establish those quantities as a compatible original-versus-revised pair. CERC paragraph 29 also reproduces text calling ₹5,566.86 million per year lower than a ₹4,785.52 million levelized tariff even though the quoted figure is numerically higher; adjacent paragraphs describe it as 16.33% higher. The inconsistency remains visible and is not converted into a project-cost conclusion.

The CERC order is an attributable regulatory record that the Share Purchase Agreement and Transmission Service Agreement were executed and the SPV was acquired. It is not the signed instrument text. The signed agreements, complete amendments, acquisition consideration, final transmission-licence order, package-level EPC contracts, payments, disputes, audits, tariff recovery, actual cost, and final cost remain unlinked.

CEA names four monitored components: the South Olpad GIS substation, the Vadodara–South Olpad and Ahmedabad–South Olpad 765 kV lines, and the Gandhar–Hazira 400 kV line-in line-out. Their MVA or circuit-kilometre scope, targets, equipment, foundation, erection, and location counts remain literal component fields. They are not averaged into a new whole-project percentage. CEA's original October 2026 and anticipated October 2027 targets are not relabeled as an executed contractual extension, approved delay finding, actual completion, or present status.

The sources publish no alignment, substation coordinate, right-of-way geometry, or surveyed project extent. The existing Olpad point remains an unchanged representative corridor navigation anchor, not South Olpad GIS or any transmission line. A newer authoritative monitoring source is required for conditions after May 2026.

Release 54 grid breadth and component-monitoring boundary

The public catalog now contains 393 source-backed records: 269 Major projects or mine assets and 124 optional Facilities or local schemes. The default non-toll browse scope contains 304 records; the same 89 reviewed toll plazas remain available only through the explicit Toll plazas layer. Twenty additional Power Grid & Storage systems extend the cumulative PAIMANA lane to 245 records while preserving every earlier identity, coordinate, precision label, source field, and individual unclustered feature.

The new records use MoSPI's May 2026 PAIMANA table for published identity, scope, state, cost, schedule, expenditure, progress, reporting agency, and reporting cutoff. Their map positions remain reviewed navigation evidence rather than project geometry. Vtaman and Bhadla were disambiguated against their named village and tehsil context; one candidate whose cached Rajasthan anchor conflicted with its Uttar Pradesh source-state field was excluded rather than silently relocated.

Koppal–Gadag now links an official Central Electricity Authority report for the status of under-construction transmission schemes as of 31 May 2026. The reviewed row publishes ₹4,445 crore as the scheme's Estimated cost, December 2023 as the special-purpose vehicle transfer month, December 2025 as the Original SCOD, and September 2026 as the Anticipated SCOD. These values align with the compatible PAIMANA scheme snapshot but remain reporting fields, not actual or final cost, payments, or proof of whole-scheme completion.

The same CEA row names seven monitored components: two new substations, three transmission lines, and two augmentation works. Their target months, circuit-kilometre or MVA scope, construction indicators, and remarks remain literal component evidence. A component indicator is not converted into a synthetic percentage, package award, contract, payment, cause of movement, or whole-project completion statement. Timeline loads these component records only when opened; the initial Project Brief remains compact.

The independent state and district checks cover all 393 points. Three hundred eighty-seven agree with their declared source-state scope, while the same six earlier NHAI conflicts remain explicit. Every point resolves to one district across 259 distinct Local Government Directory district-code pairs. The CEA report does not publish surveyed line geometry, substation coordinates, contractor packages, payment records, amendments, disputes, audit findings, or final outturn, so those fields remain missing.

Release 53 KLI agreement and compact-Overview boundary

Release 53 preserves the 373-record catalog, 284-record default non-toll scope, 89 optional toll plazas, every source coordinate, and every spatial-precision label. It deepens the Kochi–Lakshadweep Islands submarine OFC parent system with two fingerprinted official documents: the 6 January 2022 implementation agreement and the tripartite O&M agreement signed on 5 February 2024 with a stated effective date of 10 July 2023.

The implementation agreement publishes ₹810.27 crore as Total cost to be funded by USOF, exclusive of taxes and statutory levies. The O&M agreement separately publishes ₹51.05 crore as an Initial gross annual O&M charges estimate, exclusive of tax, says the heads are tentative and payable on actuals, and applies an additional PMC rule. These quantities remain separate from DBN's ₹837 crore capital-expenditure field, ₹1,072 crore Total financial implication, ₹235 crore five-year OPEX, claims, disbursements, payments, actual outturn, actual cost, and final cost.

The selected-project Overview now uses deterministic, type-aware composition. It shows a concise Project Brief, applicable facts, one source-backed important change when one exists, a compact milestone summary, exact organization roles, evidence freshness, missing evidence, and map precision. Mine assets do not receive generic construction-cost or deadline rows; inaugurated events retain their date and scope separately from completion, commissioning, opening, and present operation. Missing values stay labeled Not published or Unknown rather than becoming zero or a favorable state.

The initial brief response omits packages, payments, tenders, fee-agency contracts, shared toll revenue, and other specialist payloads. Opening Timeline, Money, Organizations, Packages, Contracts, or Evidence retrieves that section independently. This is progressive disclosure only: the underlying evidence, source roles, money definitions, schedule meanings, spatial records, missing-data states, and retry boundaries remain available.

The agreements do not establish a complete claim or payment ledger, audited annual O&M outturn, actual or final project cost, authoritative actual completion date, revised deadline, current network performance, landing-station coordinate set, or surveyed cable geometry. Kavaratti therefore remains a representative corridor navigation point, and no rollout-delay or operating-performance conclusion is inferred.

Release 52 CANI agreement and amendment boundary

Release 52 preserves the 373-record catalog, the 284-record default non-toll scope, and every source-backed coordinate. It deepens only the Chennai–Andaman and Nicobar Islands submarine OFC parent system. The record now links four official, page-reviewed documents from Digital Bharat Nidhi: the 12 July 2018 implementation agreement, 6 November 2019 fund-release amendment, 13 November 2020 tripartite O&M agreement, and 27 January 2026 O&M addendum.

The money fields remain source-specific. Clause 5.2 of the 2018 agreement publishes ₹929.358 crore as Total CAPEX cost to be funded by USOF. The 2020 O&M agreement separately publishes an initial gross annual O&M estimate of ₹49.84 crore and restricts the UT Administration's fixed annual obligation to ₹46.84 crore, with actual-based claims and an additional PMC rule. These are not added to the later DBN page's ₹1,224 crore Total project cost, silently compared with its ₹947.283 crore capital-expenditure label, or relabeled as disbursement, payment, expenditure, actual cost, or final cost.

The 2019 amendment changes clauses 5.5, 5.6, and 5.7. It makes later release contingent on physical progress, utilization evidence, purchase-order-linked claims, and proof of vendor payment, and changes the PMC and release-timing mechanics. A requirement to submit payment proof is not evidence that the payment occurred. The product therefore keeps the payment ledger missing.

The 2020 tripartite agreement assigns BSNL O&M responsibility, the UT Administration a net-OPEX viability-gap-funding role, and Pr. CCA Tamil Nadu asset and fund-management functions. The 2026 addendum states an effective date of 1 October 2025 and extends the principal O&M agreement through 30 September 2030. The extension does not establish a new annual ceiling, present network availability, maintenance quality, or service performance.

Each PDF retains the direct official URL, reviewed pages, retrieval date, and SHA-256 fingerprint. Those documents support the contract and amendment trail but do not upgrade the map point. CANI remains a representative Port Blair locality anchor, not a landing station, surveyed site, route point, cable LineString, or project extent. The staged D1 model stores the two agreements, two amendments, two explicit amendment relationships, four source documents, and exact organization roles while the public product remains on the validated static read path.

Release 51 Digital Bharat Nidhi network boundary

The public catalog now contains 373 source-backed records: 249 Major projects or mine assets and 124 optional Facilities or local schemes. The default non-toll browse scope contains 284 records; the same 89 reviewed toll plazas remain available only through the explicit Toll plazas layer. Release 51 activates Digital & Telecom with two physical parent-network records: the Chennai–Andaman and Nicobar Islands submarine OFC system and the Kochi–Lakshadweep Islands submarine OFC system.

Digital Bharat Nidhi, Department of Telecommunications, and Press Information Bureau sources establish the project identities, named island scope, approval and lifecycle events, project completion fields, exact organization roles, reported cost definitions, and inauguration evidence. CANI keeps ₹1,224 crore as Total project cost. KLI keeps ₹1,072 crore as Total financial implication (excluding taxes), while the later PIB “plus taxes” wording remains visible as a source difference. Capital expenditure, multi-year operational expenditure, total project cost, payment, expenditure, actual cost, and final cost are not collapsed.

The official sources do not publish route LineStrings or a reviewed landing-station coordinate register. CANI therefore uses Port Blair, now named Sri Vijaya Puram in the cached OpenStreetMap result, as a representative corridor navigation anchor. KLI uses Kavaratti on the same basis. Both markers retain `Corridor representative point` precision, source coordinates, WGS 84, one-time cached lookup provenance, and an explicit statement that the point is not a landing station, marine route, midpoint, surveyed site, or project extent. No cable alignment is invented.

Inauguration remains separate from completion, commissioning and current operation. CANI's 10 August 2020 event and KLI's 3 January 2024 event appear only as sourced lifecycle entries. KLI's May 2023 target is not compared mechanically with the inauguration date because the reviewed sources do not publish an authoritative actual construction-completion date or revised deadline. Present-day network availability, maintenance condition, and service quality remain missing.

The independent state and district checks now cover all 373 points. Three hundred sixty-seven points agree with their declared source-state scope, while the same six earlier NHAI conflicts remain explicit. Every point returns exactly one district across 251 distinct Local Government Directory district-code pairs; the two new locality anchors pass their expected island-territory checks without exporting boundary geometry.

Release 50 basemap, category, and lifecycle boundary

The interactive map now uses the complete OpenFreeMap Positron style without Infra's earlier blanket symbol filtering, custom warm background and water overrides, or translucent India surface. This restores the same provider palette and native zoom-dependent hierarchy used by Tech Atlas. Only the provider's country-label layers are narrowed to India so large neighboring-country names do not compete with the national subject; city, town, road, transport, water, land-cover, and built-environment layers remain native. National zoom remains calm, and those orientation layers appear progressively as their provider zoom thresholds allow. The source-backed India outline and state boundaries remain independent official orientation overlays, and project layers remain above the basemap.

Provider features are orientation only. A basemap road, railway, airport, station, waterway, building, place, or land-cover feature does not establish a monitored project, construction state, project geometry, cost, deadline, contractor, progress, or evidence tier. If provider tiles, glyphs, sprites, or labels fail, the compact spatial index, directory, selection, Project Brief, specialist evidence, official geography retry, and map-failure access remain separate.

The maintained physical-infrastructure vocabulary now has 18 categories: Highway / Road, Railway, Metro, Airport, Water Resources, Energy Generation, Water System, Port, Industrial Corridor, Mine, Urban Infrastructure, Digital & Telecom, Power Grid & Storage, Pipelines & Fuel, Logistics & Inland Transport, Environmental Utilities, Shipbuilding & Ship Repair, and Space Launch Infrastructure. Existing records migrate through deterministic legacy aliases plus reviewed sector and source-type rules. Bridges and tunnels remain with their supported transport system; hydropower generation stays separate from irrigation and water resources; generation stays separate from grid and storage; ports stay separate from inland logistics; water supply stays separate from sewerage and other environmental utilities; shipbuilding facilities stay separate from port assets and vessel contracts; and public launch facilities stay separate from sensitive range geometry and mission-readiness claims. Digital & Telecom activated only after reviewed Release 51 evidence, Shipbuilding & Ship Repair after the bounded Release 104 review, and Space Launch Infrastructure after the bounded Release 105 public-source, safety, identity, relationship, and rendering review.

Inaugurated is a source-scoped lifecycle event, not an infrastructure category. It appears only when an attributable official source explicitly publishes the event. The record preserves event date, published scope, source and fingerprint, reporting cutoff, interpretation boundary, and inaugurated-by identity only when stated. It is never inferred to mean construction completion, commissioning, opening to traffic or the public, operation of the full project, or 100% progress.

Kanpur Metro demonstrates the distinction. UPMRC's Annual Report 2024-25 says the 9 km priority corridor was inaugurated on 28 December 2021 and the Motijheel to Kanpur Central passenger-service extension was inaugurated on 30 May 2025. The same printed page says the latter brought the IIT Kanpur to Kanpur Central stretch to 14 operational stations and that construction on remaining sections was ongoing. The separate May 2026 PAIMANA row continues to supply full-project progress, cost, schedule, reporting role, and reporting cutoff.

The default browse scope contains 282 non-toll records so major infrastructure, systems, mine assets, and urban-service records are not visually dominated by fee-plaza facilities. The 89 reviewed Toll plaza entities remain unchanged inside the 371-record catalog. An explicit Toll plazas toggle adds that secondary layer to both map and results and persists as tolls=1 in shared URLs. The toggle composes with search, categories, lifecycle, state, evidence, awareness questions, and other filters. A named toll search reports the hidden matching count and offers an enable-layer action rather than returning a misleading empty catalog; a direct project URL continues to open the toll dossier and selected source coordinate without changing the stored default.

Release 49 coverage boundary

The public catalog now contains 371 source-backed records: 247 Major projects or mine assets and 124 optional Facilities or local schemes. Release 49 preserves every earlier record and the exact first 200 PAIMANA selections, then adds a ninth reviewed batch of 25 Major projects from the same 1,987-row May 2026 PAIMANA source universe. The cumulative PAIMANA lane now covers 225 records across 34 state or union-territory scopes and 20 official sectors.

The added batch spans metropolitan sewerage, three national public-institution campus programmes, an industrial node, energy terminals and pipelines, renewable-energy transmission, five intercity rail corridors, four metro systems, three national-highway corridors, hydropower, and coal development. Selection uses official scope, geography, capacity or length, corridor role, public function, regional balance, and spatial-review readiness, not cost alone.

PAIMANA publishes no project coordinates or geometry for these 225 records. Five separately reviewed official sources publish project or dam-location reference points, and one official coordinate range supports a documented representative derivation. The other 219 positions remain visibly and textually representative facility, locality, or corridor context. Every spatial source, retrieval date, original coordinate, WGS 84 reference, transformation, validation, and limitation remains attributable.

The independent state and district checks cover all 371 points. Three hundred sixty-five points agree with their declared source-state scope, while the same six earlier NHAI conflicts remain explicit. Every point returns exactly one district, spanning 250 distinct Local Government Directory district-code pairs, without downloading or republishing state or district geometry.

The reviewed calendar schedules the June 2026 Flash Report for 25 July 2026. The May 2026 report therefore remains the newest published national PAIMANA breadth source used for this 16 July release. A newer report must be retrieved, audited, and semantically reviewed before any later status is claimed.

Release 48 marker and orientation boundary

Release 48 changes presentation only. It preserves all 346 records, record IDs, source coordinates, WGS 84 reference, spatial types, precision labels, categories, evidence tiers, source roles, financial and schedule meanings, filters, lenses, search, URLs, selection, Project Briefs, specialist evidence, and map-failure access.

Release 48 introduced compact circular medallions with separate category, precision, metric, selection, and hit-area channels. Release 75 later superseded only that enclosing visual treatment with freestanding pictograms. Stable Major-first ordering, larger invisible hit targets, and separate active layers continue to keep selected, focused, and hovered records legible without DOM markers, clustering, count markers, coordinate jitter, or positional transitions.

Spatial precision remains independent of category. Release 75 demotes the earlier enclosing precision rings to short solid, double, dotted, tailed, mine-representative, or unresolved attached marks. These decorations are deliberately secondary. The map key, accessible result list, preview, Project Brief, and evidence record state the spatial meaning directly.

The OpenFreeMap Positron style now retains a reviewed set of provider water, highway-name, non-US road-shield, and airport symbols instead of removing every symbol layer. Official Indian state and city context remains source-backed and India-filtered. Regional rail orientation, towns, villages, neighbourhoods, local roads, airports, waterways, land cover, and buildings reveal progressively at their supported zooms. Low-value surrounding-country labels remain suppressed, and India's official outline stays the strongest national boundary.

Provider context is orientation only. An OpenStreetMap or OpenMapTiles road, railway, airport, station, waterway, building, land-cover feature, or place label does not establish an infrastructure record, project scope, construction status, project geometry, cost, deadline, contractor, progress, or evidence tier. Real project points and official India geography remain available before optional contextual symbols settle. Search, results, project selection, dossiers, and evidence do not depend on provider glyph, sprite, icon, or tile success.

Release 47 Chennai Metro Phase II contract boundary

The Chennai Metro Phase II dossier joins the existing curated PAIMANA record to official CMRL sources using the project identity and implementing authority. PAIMANA remains the source for 53% reported progress, ₹63,246 crore Original Cost, the same ₹63,246 crore current reported cost, and the August 2029 reported completion in the April 2026 snapshot. CMRL sources separately add the 118.9 km project scope, consultant and funding roles, package counts, award-register rows, and signed-contract evidence. One source is not silently used to redefine another.

CMRL's project-status page, updated 24 March 2026, reports 60 packages in total: 19 civil, five track, and 36 systems. It reports all civil and track packages awarded, 34 systems packages awarded, and two systems packages under finalisation. The dossier's 28 reviewed rows are a bounded evidence-linked subset, not a complete package register, and aggregate counts do not supply missing package-specific firms, dates, values, payments, amendments, or completion states.

Published package values use different inclusion bases. Some include provisional sums, GST, or customs duty; others exclude GST or describe taxes and duties differently. The interface keeps each amount with its source definition and does not calculate a package-row sum, project award total, expenditure total, actual cost, or final cost.

The archived register prints two values for ASA07: ₹91,85,06,227.23 in one table and ₹89,85,06,227.23 in a later detail block. The ₹2 crore difference remains unresolved because no reviewed CMRL source labels one value a revision, amendment, variation, or superseding amount.

Two separate CMRL releases establish the ₹1,620 crore ASA-04 signalling agreement and the ₹1,817.54 crore C5-CP06-UG06 underground-stations and tunnels agreement. Their named counterparties and scopes remain exact. They are not relabeled as one whole-project contractor, payment, project cost, actual cost, final cost, or proof of present-day performance. The map remains a Chennai locality representative point rather than official corridor, station, tunnel, depot, or work-site geometry.

Release 46 MAHSR contract boundary

The Mumbai–Ahmedabad High Speed Rail dossier joins the existing 508 km PAIMANA record to official NHSRCL pages using the corridor identity and implementing authority. PAIMANA remains the source for 60% reported progress, ₹1,08,000 crore Original Cost, the same ₹1,08,000 crore current reported cost, and the December 2029 reported completion in the April 2026 snapshot. NHSRCL sources add package structure, dated award or agreement events, exact package and consultant roles, and published contract values. One source is not silently used to redefine another.

The reviewed package slice contains 17 rows: all 11 civil package identities, P4X and P4Y steel-bridge packages, T2 and T3 track packages, D2 Sabarmati Depot, and TI1 training-institute works. NHSRCL's 20 July 2023 release says all 11 civil packages had been awarded. That aggregate statement does not supply every package-specific contractor, date, agreement, or amount, so those fields remain missing where a direct reviewed row was not located. Bidder, finance-bid, and technical-bid pages are not used as proof of award.

Several awarded-register rows publish separate INR, JPY, and USD components. Those native values remain separate and no exchange-rate conversion or combined award total is calculated. T3 also has an approximate ₹3,141 crore value in a dated press release and a separate INR-plus-JPY register presentation. The reviewed sources do not define their relationship, so no difference, revision, or overrun is calculated.

The awarded register contains two D2 rows for the same Sojitz–Larsen & Toubro consortium with ₹1,589.027004924 crore and ₹2,180.7823045 crore values, a difference of ₹591.755299576 crore. No reviewed source labels one a revision, amendment, variation, or final value. Both remain visible as an unresolved source difference rather than selecting a preferred amount.

Three additional retrieval-dated awarded-register records cover civil-works project management consultancy, signalling and telecommunications consultancy, and the S1 signalling, telecommunications, and operational-control-centre systems contract. Their named organizations and values remain attached to those exact roles. They are not relabeled as one whole-project contractor, project expenditure, payment, actual cost, or final cost. Package payments, amendments, disputes, audits, final costs, and official corridor or package geometry remain incomplete.

Release 45 Mumbai Metro Line 3 contract boundary

The Mumbai Metro Line 3 dossier joins PAIMANA project code 702637 to official MMRCL pages that identify the same project and implementing entity. The PAIMANA snapshot remains the source for 97.86% physical progress, ₹23,136 crore Original Cost, ₹37,276 crore Revised Cost, ₹33,648.26 crore Cumulative Expenditure, and the March 2023 to August 2025 schedule movement. MMRCL's pages supply package rows, named parties, work-order terms, a funding plan, and the General Consultant role. One source is not silently used to redefine another.

MMRCL says the project has 18 main packages, while its table displays 22 rows because Packages 10, 11, and 16 contain lettered subpackages or lots. Nineteen rows show an award or Letter of Acceptance state. Package 18 says General Consultants engaged. Package 9 and Package 13 publish neither a contractor nor an award outcome in the displayed row. These counts are reported separately.

Only 13 of the 22 package rows publish amounts. Their ₹19,640.03 crore sum is calculated solely to describe those displayed rows. It is not total awarded value across all packages, project cost, cumulative expenditure, payment, actual cost, or final cost and is not compared with PAIMANA project cost as an overrun. The six operations-related amounts are work-order amounts for separate service scopes and remain separate from both package amounts and project cost.

MMRCL's funding page publishes ₹23,136 crore as Total Project Completion Cost. Its eight displayed component amounts sum to ₹23,137 crore, while the printed percentages sum to 100%. The product preserves the ₹1 crore rounding or source difference instead of changing a component or the published total. The page has no visible update date, so this remains a retrieval-dated funding-plan record, not a current revised-cost, disbursement, actual-cost, or final-cost statement.

Only the AFC Package 14 press release supplies a reviewed award date, 15 January 2019. The other package award dates and underlying signed documents remain unlinked. The operations table does not establish current performance, payments, renewals, amendments, or service status after each published tenure. The representative Mumbai marker remains a corridor navigation point; no station, tunnel alignment, depot boundary, or complete project geometry is invented.

Release 44 evidence-question boundary

Six suggested questions are deterministic views over the same reviewed current facts. They are not live AI answers and do not introduce project data. Each result exposes the exact source-bounded reason it matched. The summary reports the question definition, records in scope, records with the required compatible fields, missing-data count, matching count, and catalog-generation date.

Airports in progress requires the reviewed Airports category, a source-published physical-progress value above 0% and below 100%, and the record's reporting cutoff. High-speed rail currently matches only Mumbai–Ahmedabad because reviewed official evidence explicitly establishes that identity. Other rail records are not classified from cost or a loose name match.

Past reported deadline compares a central project's current or revised completion date with that record's own status cutoff and requires progress below 100% in the same snapshot. The 245 PAIMANA breadth records use 24 June 2026. The earlier curated central-project snapshot uses 30 April 2026. Records without a completion date, progress, or reporting cutoff remain visible in the missing-data denominator.

Ten-plus-year extension compares only compatible original and current or revised completion dates. Cost at least two times compares only compatible original and revised reported project costs, never grants, awards, expenditure, payments, tender values, revenue, or final cost. Needs newer evidence matches only an explicit published freshness or review warning; no warning is not a certification of freshness or completeness.

The current catalog returns 17, 1, 26, 5, 17, and 29 matches respectively. These results are not findings about cause, fault, corruption, waste, success, current operation, or present-day status. Filters can narrow a question, and the denominator and missing-data count recompute over the filtered scope. The question identifier, lens, filters, selection, and camera remain shareable in the URL.

An unthrottled cache-disabled local 390 by 844 check showed the national overview at 24.5 ms, browse controls at 63.0 ms, all 346 native project features at 160.7 ms, orientation geography at 226.9 ms, FCP and LCP at 56 ms, zero layout shift, no long task, and about 33.3 MB used JS heap. The comparable Release 43 single run recorded 18.6, 74.8, 119.8, and 194.3 ms, FCP and LCP at 72 ms, and about 61.3 MB. A 1.4-second map-drag sample reported 60.1 frames per second, 17.7 ms p95 frame time, and no frame over 50 ms. Twenty sequential warm local question API requests measured 1.3 ms p95. These are controlled local observations, not production percentiles or evidence that the public latency budgets are met.

Release 41 startup delivery boundary

Release 41 changes when reviewed geography becomes visible. It does not alter any record identity, category, source coordinate, spatial precision, role, money, schedule, evidence, attention signal, Project Brief, specialist view, filter, URL, or map-failure fallback.

A deterministic build generates a non-interactive national loading overview from the current spatial index, reviewed India outline, and reviewed state boundaries. It contains one feature for each of the 346 published records, preserves every longitude and latitude exactly, uses category color and precision shape, and keeps Major projects or mine assets more prominent than Facilities or local schemes. The release gate checks the record count, unique IDs, unchanged coordinates, data version, and SHA-256 hashes of the spatial and orientation inputs. Clustering, count bubbles, random sampling, coordinate jitter, and invented displacement are rejected.

The overview is orientation during startup, not the authoritative interactive surface. It cannot be hovered, selected, zoomed, or used to infer a project extent. MapLibre remains responsible for navigation, selection, precision explanation, and linked evidence. A visible loading label states that the interactive map is still being prepared.

The exact MapLibre GL JS 5.24.0 browser distribution is copied to a stable public path during the build and validated against its version, byte count, BSD-3-Clause license, and SHA-256 hash. The browser can reuse that low-priority preload instead of discovering a separate renderer chunk after application hydration. The initial OpenFreeMap style omits provider symbol layers so unrelated glyph requests do not gate reviewed project features. Curated Indian state, capital, major-city, and progressive local context still installs from the separately validated geography package.

Under a controlled cache-disabled profile at 390 by 844 with 150 ms latency, 1.6 Mbps download, and 4-times CPU throttling, the Release 40 baseline showed first native project features at 22.44 seconds, browse controls at 9.40 seconds, orientation geography at 25.45 seconds, and long tasks of 215 ms and 56 ms. Release 41 showed the truthful overview at 3.73 seconds, browse controls at 8.66 seconds, all 296 native project features at 11.70 seconds, orientation geography at 14.88 seconds, and no long task. An unthrottled 1,440 by 900 drag and wheel sample reported 60.7 frames per second, 17.6 ms p95 frame time, and no frame over 50 ms. These are controlled local observations, not production p75 values.

The constrained interactive time still exceeds the aspirational 3-second mobile target. The renderer remains the largest first-use asset, and production field data is not yet available. Later work must continue reducing startup cost while preserving direct gestures, individual coordinate-locked features, accessibility, and map-independent evidence access.

Release 43 coverage boundary

The public catalog now contains 346 source-backed records: 222 Major projects or mine assets and 124 optional Facilities or local schemes. Release 43 preserves every earlier record and adds 25 reviewed Major projects from the same 1,987-row PAIMANA May 2026 source universe. The cumulative PAIMANA lane now covers 200 records across 34 state or union-territory scopes and 20 official sectors.

The added batch spans airports, metro systems, rail corridors, a Brahmaputra bridge, national highways, hydro and power systems, transmission and energy pipelines, major water systems, a national public-university campus phase, and coal development. Selection uses official system scope, geography, capacity or length, corridor role, and public function, not cost alone.

PAIMANA publishes project identity, reporting roles, financial fields, dates, and physical progress but no project coordinates or geometry. Five separately reviewed official documents now publish a project or dam-location point for Sunni Dam, Rangit IV, Dhaulasidh, Heo, and Kiru; each remains an official reference point rather than a surveyed component, boundary, parcel, or complete extent. Tirap uses a documented representative point derived from an official project-area coordinate range. It is not presented as a source-published point, lease centroid, mine boundary, extraction footprint, or current operating-state claim. The other 194 PAIMANA positions remain visibly and textually approximate. Additional official publications may support only named context, including Marrivalasa for the Central Tribal University permanent-campus area, without importing geometry or unrelated lifecycle, procurement, or financial claims. The pipeline retains each spatial source, retrieval date, original coordinates, coordinate reference system, transformation, validation, and limitation.

The independent state and district checks now cover all 346 points. Three hundred forty points agree with their declared source-state scope, while the same six earlier NHAI conflicts remain explicit. Every point returns exactly one district, spanning 237 distinct Local Government Directory district-code pairs, without downloading or republishing state or district geometry.

Under the same controlled cache-disabled 390 by 844, 150 ms latency, 1.6 Mbps, and 4-times CPU profile, Release 43 showed the truthful 346-record overview at 3.40 seconds, browse controls at 9.51 seconds, all native project features at 12.16 seconds, orientation geography at 15.34 seconds, FCP and LCP at 3.40 seconds, zero layout shift, and no long task. The comparable Release 42 values for 321 records were 3.40, 9.12, 11.95, and 15.14 seconds. An unthrottled cache-disabled run showed the overview at 18.6 ms, browse controls at 74.8 ms, all 346 project features at 119.8 ms, orientation geography at 194.3 ms, FCP and LCP at 72 ms, zero layout shift, no long task, and about 61.3 MB used JS heap. A 1.2-second map-drag sample reported 61.2 fps, 17.5 ms p95 frame time, and no frame over 50 ms. The constrained Tirap Project Brief became visible in 3.06 seconds; its brief API request took 178 ms. These are controlled local observations, not production p75 values. The constrained interactive feature and geography times remain outside the aspirational budgets.

Release 93 compact activation boundary

The catalog remains the same 10,093 reviewed records. When the user is viewing the complete All records scope without search or membership-changing filters, the discovery response may state `all-non-toll` or `all-records` instead of serializing thousands of record IDs that are already present in the validated spatial index. The browser applies that exact scope locally. Every matching project remains an individual source-locked MapLibre feature.

Compact activation is not used for category, state, district, lifecycle, evidence, freshness, spatial precision, organization, attention, mining, awareness, search, or Major and Facilities filters. Those states retain an explicit ordered ID set from the same version-gated D1 query. A request that does not opt into compact activation also retains the earlier complete-ID response contract.

The spatial index is authoritative only for this narrow activation decision: include every published record, or include every published non-toll record. It does not independently reproduce richer database filters or upgrade record meaning. Counts, facets, cursor pages, match reasons, missing-data counts, toll disclosure, selected detail, evidence, and static failure recovery remain unchanged.

This release does not claim general viewport querying or a million-record map solution. Broad filtered states may still return many explicit IDs, and the complete compact spatial index still loads before map activation. Later work must bound those paths without clustering, sampling, moving coordinates, hiding accessible records, or weakening evidence semantics.

Release 92 D1 discovery boundary

The catalog remains the same 10,093 reviewed records, including 10,004 core non-toll records and 89 separately reported toll-plaza facilities. Release 92 changes only the healthy filtered-discovery read path. It does not alter identity, category, coordinates, precision, lifecycle, evidence, money, schedule, organizations, markers, dossiers, default toll behavior, or URL semantics.

A versioned D1 presentation projection stores only the fields needed to filter, sort, explain, and render concise browse rows. Category, state, lifecycle, awareness, toll, precision, mining, evidence, and cursor queries are checked for exact semantic parity with the validated static current-facts model. The projection also stores the deterministic JavaScript name-sort ordinal so the database cannot silently reorder equal or punctuation-heavy names.

D1 is used only when the bound read model reports the exact public data version. A missing binding, stale projection, or failed query is classified and returns the independently validated static discovery response. Response headers expose whether D1 or static fallback served the request without revealing internal error details. Selected Project Briefs, specialist evidence shards, the spatial index, and the map-failure directory remain separate reads.

The migration repairs one previously missing normalized state relationship for a multi-state NHAI road record. That repair restores filter parity without changing the official source wording, canonical identity, coordinates, representative precision, or evidence. Idempotent state inserts prevent a rerun from duplicating the relationship.

This is not a full-text-search, viewport-query, vector-tile, or production-percentile claim. The healthy path no longer fetches the 23.9 MB current-facts artifact merely to answer a filtered request, but it still returns the complete ordered active-ID set so every current match remains individually available to the unclustered map. Larger catalogs will require bounded viewport activation and a reviewed search index.

Release 36 discovery delivery boundary

The public catalog still contains the same 196 source-backed records. Release 36 changes how browse results and facets reach the browser. It does not alter project identity, category, source coordinates, spatial precision, evidence tier, role, money, schedule, freshness, attention, Project Brief, specialist evidence, or URL semantics.

The compact spatial index still arrives first and activates every published record as an individual MapLibre feature. The browse workspace requests 24 results at a time from a versioned cursor API. Each response also carries the exact matching count, complete matching IDs for the current 346-record map, deterministic category counts and filter options, missing-data count for the active lens, catalog-generation date, and one concise reason each row matched.

Cursor order is deterministic and ends with stable record ID as a tie-breaker. A cursor is bound to the exact search, filter, layer, and lens state, so it cannot be reused against a different result set. The release validator traverses every page and checks that all 346 records appear once without overlap or omission.

The complete current-facts artifact no longer loads during the healthy first browse. It remains a separately validated static fallback if the discovery API is unavailable, preserving map-independent search and filter access instead of presenting an empty directory. The compact first page is tested as a separate payload rather than described as a production p75 latency claim.

A cached local production run at 1,440 by 900 CSS pixels reported the spatial index at 54.6 ms, browse controls at 57.2 ms, all 296 project markers at 386.1 ms, orientation geography at 458.8 ms, FCP and LCP at 88 ms, zero layout shift, and no long task. A 1.2-second map-drag sample reported 61.2 frames per second, 18.1 ms p95 frame time, and no frame over 50 ms. These are single-machine observations, not field percentiles.

The Release 40 constrained baseline remained over budget, with first native features at 22.44 seconds. Release 41 reduces that to 11.70 seconds and supplies the 296-record source-backed overview at 3.73 seconds, but the interactive path still does not meet the aspirational mobile loading budget. Search, records, and the map-failure directory remain usable. Further renderer, spatial delivery, and production-cache work is required before claiming mobile performance compliance.

This is not yet a production D1 query path or a million-record spatial activation claim. The API currently reads the validated release artifact on the server and returns every matching ID because the complete catalog is 346 records. D1, full-text search, cursor-backed database queries, viewport activation, vector tiles, production concurrency evidence, and immutable hosting caches remain staged.

Release 35 geography delivery boundary

The public catalog still contains the same 196 source-backed records. Release 35 changes how the reviewed India outline, state boundaries, and state labels reach the browser. It does not alter their coordinates, features, labels, coordinate reference system, derivation, simplification, provenance, precision, or limitations.

A deterministic build writes three content-addressed gzip assets from the committed reviewed GeoJSON sources. The release gate verifies each source SHA-256 hash, raw and compressed byte count, feature count, decompression, and deep GeoJSON equality. The package is therefore byte-for-byte equivalent after decompression at the JSON-data level, while reducing the combined geography transfer from 1,604,961 bytes to 504,470 bytes.

The browser loads the small versioned manifest and the three compressed assets in parallel. If the manifest, browser decompression stream, compressed request, or decompression step is unavailable, that asset falls back to the same reviewed raw GeoJSON. If both optimized and raw paths fail, only the official orientation layer becomes unavailable; the basemap, individual project points, search, directory, Project Briefs, specialist evidence, and retry path remain independent.

The curated national and regional city sets install after the map and real project points. Denser Indian town and village layers remain clipped to the reviewed outline and are installed only when the reader reaches their supported zoom. This avoids compiling close-zoom orientation filters during the national first read while preserving progressive context during exploration.

This delivery package is not a vector-tile claim. PMTiles, production D1 and R2 reads, immutable hosting headers, and larger-scale viewport delivery remain staged work.

Release 34 evidence delivery boundary

The public catalog still contains the same 196 source-backed records. Release 34 changes how specialist evidence is delivered, not its facts, visual hierarchy, coordinates, evidence, or discovery behavior.

The compact spatial index still loads before current discovery facts. A selection now requests the evidence-checked Project Brief view first. Its TL;DR, current facts, compatible money and schedule movement, exact organization roles, procurement summary, freshness, gaps, source identity, and spatial precision remain available without retrieving specialist explanatory prose.

Organization-role families are discovery aids derived from published role headlines. Project Briefs and Evidence keep the exact wording; a family does not establish ownership, present operation, blame, or misconduct.

Opening Timeline, Money, Organizations, Packages, Contracts, or Evidence now requests an independent, cacheable response for that named section. Each successfully loaded section is reused when the reader returns to it. Loading, unavailable, retry, and return-to-Overview states remain isolated per section, so one failure does not remove the Project Brief, another cached section, record identity, map, directory, or evidence boundary.

The earlier full investigation view remains available as a compatibility response, but the public interface no longer requests it. Evidence is still the largest section because it must restore source fields across nested contracts, payments, disputes, and packages. Further DTO normalization can reduce that response without weakening source identity.

All 196 active records render as individual MapLibre features at their unchanged published coordinates. A lightweight native point source paints independently of the richer precision and category layers, so real locations remain the first spatial result. Native layers stay locked to the basemap during drag, wheel, pinch, and animated camera movement. Hover and selection use feature state rather than recreating every marker. The result list remains the equivalent keyboard path and keeps the same preview information.

Marker meaning uses separate channels. The freestanding professional pictogram and its stable color identify the reviewed infrastructure category. A small attached mark identifies exact, official-facility, locality, corridor, mine-representative, or unresolved placement. A metric lens may add a separate semantic ring, and a silhouette-following neutral outline identifies selection or focus. Color reinforces these meanings but does not replace the pictogram, precision mark, label, or accessible text.

A 22-table D1 schema, immutable-artifact references for R2 and PMTiles, an idempotent 346-record import, six materialized awareness states per record, and revision checkpoints are prepared but are not the production read path. The Site has no D1 or R2 binding in this release. A switch requires semantic-equivalence checks, production-like query and concurrency evidence, cache review, resilience testing, and exact deployment verification.

The included 100,000, 1 million, and 5 million row test is a local SQLite metadata capability test, not a production D1 latency claim. It does not test full-text search, tiles, concurrency, network latency, or R2. Separately partitioned specialist views, production D1 reads, immutable R2 artifacts, and tiled geography remain staged work.

1. Published facts

Project progress, reported costs, completion dates, highway identifiers, operation dates, and named organizations are published only when a linked source supports the field.

2. Derived signals

Cost change is calculated as revised cost minus original cost within the same project record. A “past reported completion” signal means the revised date had passed by the April 2026 PAIMANA snapshot while reported progress remained below 100%. A toll fee freshness signal means the fee-revision due date on the NHTIS page had already passed at retrieval. A tender-status review signal means the portal still showed opening in progress after its scheduled bid-opening date; it does not establish whether the procurement is delayed or the portal is unrefreshed. A fee-agency outcome signal means one or more linked solicitations lack a published selected agency or award result. It does not imply that no award occurred. A historic arbitration signal is bounded to its disclosure date. When a later court outcome is linked, the signal changes to the specific remaining evidence gap instead of carrying an old pending status forward.

AMRUT's literal Project Cost field is not entered into the original-versus-revised series because the directory does not define it as original, sanctioned, tender, award, revised, expenditure, actual, or final cost. Its milestone is displayed as a source-dated directory label, not independently verified current construction status. No cost-change, deadline, contractor, or award conclusion is generated from either field.

Mine lifecycle is also dated. Five IBM records publish production for FY 2023-24 or FY 2022-23, Gollaghat publishes an inspection-dated suspension observation, and Surda publishes a reporting-year stop followed by lease execution without establishing resumption. The interface does not carry any of those observations forward as current in 2026. A completed or in-progress mining-development project does not prove that a mine asset is operating.

These signals help a reader decide what to inspect. They do not establish failure, waste, misconduct, or corruption. Color can reinforce a written state, but never replaces its label.

3. Geographic precision

Exact toll point

Published NHAI latitude and longitude for a named fee plaza. The product stores the original values, WGS 84 reference system, chainage context, source, and transformation into longitude-latitude map order.

Official facility point

A coordinate published for a named toll-plaza facility in an official source whose list is described as indicative. It is more attributable than an editorial placement but is not presented as surveyed accuracy, especially when another official coordinate conflicts.

Facility point

A representative point for a known campus, airport, or facility complex. It is not an official survey coordinate or a construction-package boundary unless the spatial evidence says so.

Locality

A city, town, pass, or named project area. For AMRUT breadth records, a cached locality match represents the official city or ULB scope because the project page publishes no coordinates. The marker is not the project site, ward, centroid, package extent, or surveyed construction point.

Mine-asset representative

A reproducible point derived from official coordinate limits, or the least misleading named locality or mine-feature representative when official geometry is absent. Every mine record states its source, transformation, coordinate-reference limitation, and whether the point is official, derived, or OpenStreetMap context. No point is a lease centroid, cadastral boundary, mine footprint, extraction area, reserve extent, or surveyed site.

Corridor

A representative point for a linear or multi-state project. It must not be read as the only work site.

Unresolved official location

An official facility point whose state, district, corridor, or coordinate records still conflict after review. A dashed outer diamond supplements the normal precision shape, and the marker is not moved to an invented replacement point.

4. Every marker is evidence

Every mapped record preserves the original stored coordinates, coordinate reference system, spatial type, precision class, source link, retrieval date, any transformation, validation note, and known conflict. The map uses one short attached line for an exact toll point, a double line for an official facility point, a dotted line for a representative place or facility, a tailed line for a representative corridor, and a separate question mark for unresolved official geography. The attached mark carries precision even when a map lens changes color or size.

PAIMANA does not publish coordinates in the inspected project extracts. The earlier 15 records use explicitly labeled editorial navigation points derived from the source-named place or corridor. Releases 27, 28, 37, 38, and 39 publish 124 cached OpenStreetMap representatives for source-named facilities, localities, or corridors, with the exact OSM entity, retrieval date, query, and limitation retained. A separate official SJVN document supplies one project-location coordinate for Sunni Dam; it remains an official reference point rather than surveyed component geometry. Each later batch reuses the reviewed cache and retrieves only new or changed matches. None is described as official PAIMANA geometry. Seven deeper toll records use cross-checked exact points from the NHAI coordinate annexure. Patas and Sardewadi use later official IHMCL facility points, visually distinguished as indicative rather than exact, while the conflicting 2017 NHAI coordinates remain visible in each evidence trail.

Release 20 adds 80 NHAI RajmargYatra breadth records as official toll-plaza facility points. The public record supplies latitude, longitude, state, highway, chainage, corridor, PIU, and regional-office context, but it does not claim surveyed accuracy. These points therefore use the official-facility double mark rather than the exact-point single mark. Release 26 removes state counts and every other zoom-based aggregation from all map lenses. One active record always produces one precision-aware map feature.

The Major projects layer contains 6,537 reviewed projects, systems, or assets. The Facilities layer contains 4,490 separately modeled facilities or local schemes, including 89 toll plazas, 1,095 Amrit Bharat station developments, 1,053 AMRUT urban-services records, and 36 Namami Gange sewerage-project records. Classification follows reviewed official scope, authority, geography, capacity, system role, or asset-specific identity, not an interface-generated cost ranking or project-name guess. Named toll plazas remain an independent opt-in facility layer, so search covers the active non-toll scope by default and exposes the matching toll count and enable-layer action for an intentional toll query.

Release 29 introduced the project-type taxonomy as a reviewed discovery layer, not a new source fact. Seventeen categories are currently maintained, and only categories established by published records appear as active discovery options. A single record receives one primary category from its official sector plus reviewed scope. The category pictogram is repeated across filters, markers, results, previews, and the Project Brief, while the independent attached mark and text continue to show whether the position is exact, official-facility, representative locality, representative corridor, or unresolved.

Release 25 adds 35 AMRUT 2.0 basic official records, one for every state or union territory in the Ministry directory. The official pages supply a city or ULB but no project point or geometry. A one-time, sequential, cached OpenStreetMap Nominatim lookup therefore supplies a square locality representative point with the OSM entity, query, coordinate, WGS 84 order, retrieval date, attribution, state check, and limitation preserved. The production product makes no live Nominatim request, and the editorial point is never presented as an official project site.

For Patas, the official points are 4.92 km apart. For Sardewadi, they are 224.75 km apart. The later IHMCL points align with the Patas–Sardewadi–Varwade corridor sequence and are used for navigation, but that contextual fit does not turn them into surveyed coordinates or erase the conflict.

Release 22 introduced the independent state check for 104 records. The current local Release 123 snapshot covers all 11,027 candidate map positions. The service returns one containing state for 10,994 points: 10,964 agree with the project's declared source scope and 30 source-state differences remain explicit. Twenty-seven official coastal port points and six source-backed administrative label anchors return no land-polygon match. The product preserves the original coordinates, source state, returned state, per-row query date, LGD code, Census code, method, source limitation, and no-match state without moving a marker.

Release 23 reviews those six conflicts against attributable official documents. Ghamroj is listed in Haryana by a Ministry of Road Transport and Highways parliamentary annexure; the Kiranj Gazette names Nuh district, Haryana; the Koyala Gazette names Rampur district, Uttar Pradesh; and the Sayed Mazera Gazette names Saharanpur district, Uttar Pradesh. Those four reviewed states now support geographic filtering, while the earlier RajmargYatra state fields remain visible as source differences. The review does not change their coordinates or upgrade them to surveyed points.

Mudipar and Moura remain unresolved. Mudipar's NHAI corridor and fee-order evidence supports Chhattisgarh, while the published coordinate falls in Rajasthan. No authoritative replacement coordinate was located. Moura's point and Rupnagar district wording point to Punjab, while the RajmargYatra field and the same Gazette sentence say Himachal Pradesh; neither state is presented as a settled correction. Both use a dashed conflicting-location marker, remain discoverable from both state signals, and expose the source page, document hash, limitation, and full evidence link.

Release 24 introduced district containment for 104 records. The current local Release 123 snapshot applies the same returnGeometry=false check to all 11,027 points. Of these, 10,993 return exactly one district across 753 distinct Local Government Directory district-code pairs. Kakari Mahudi Dam retains two returned districts without selecting one; 33 points retain no district match. Each unambiguous district's state name and LGD state code agree with the separate state-containment result. The mapped district remains separately sourced context for the displayed point and does not replace a source locality, station identity, project scope, asset extent, corridor, cable system, city system, or State/UT programme scope.

Mapped district means the district containing the displayed point at the 16 July 2026 review. For exact and official facility points it is useful facility context. For a locality, representative facility, corridor, or multi-state project it remains context for the navigation point only. It does not establish the project's full district scope, validate a whole line or package, or turn an editorial placement into official project geometry.

The containment layer is an Esri India analysis service credited to Census of India, Survey of India, Local Government Directory, and Esri India. It is independent of the NHAI state field and the displayed 1:16M linework, but it is not a directly downloaded Survey of India polygon file. Its terms prohibit offline export, so the pipeline requests no geometry and stores only the returned administrative attributes and source metadata. Containment confirms which state polygon contains the mapped point at review time; it does not upgrade the point to surveyed accuracy or validate a corridor's full extent.

The state and union-territory context is a separate orientation layer. Its 493 line strings and 36 label anchors are derived from Survey of India 1:16M GeoPDF layers using their embedded Lambert Conformal Conic control points, transformed to WGS 84, and simplified at 0.008 degrees for web display. Source URLs, document hashes, the full source CRS, control points, transformation, vertex counts, and limitations are published with the data. This generalized layer is not project geometry, a cadastral boundary, or surveyed local evidence.

State-capital and important-major-city labels use OpenMapTiles place points from the OpenFreeMap basemap, clipped to the official India outline. The 2026 Survey of India political map is the curation reference for the national hierarchy. Additional regional cities, towns, and villages appear progressively at closer zoom. These labels support orientation only and never substitute for a project's spatial evidence. No municipal boundary is published in Release 30 because no reviewed authoritative, attributable, licensable geometry met the release contract. The Esri India district service explicitly prohibits offline export, so its polygons are not copied into the Site.

5. Map interaction follows spatial precision

The full-viewport map uses direct wheel, trackpad, drag, pinch, and keyboard-accessible controls. Pitch and rotation are constrained because they do not add evidence value. Selecting an exact point may open at a local scale, while an official facility, locality, or representative corridor opens progressively farther out. A corridor point is not treated as a corridor extent. Closing a brief restores the previous browse camera unless the user deliberately changed the camera while the brief was open. Hover and keyboard preview never move the camera.

Every record in the active Major projects, Facilities, or All records layer renders individually in every lens, including Coverage and national zoom. MapLibre native GeoJSON symbol layers own the unchanged published coordinates, while category, precision, hover, and selection remain separate visual properties. Feature state changes interaction emphasis without rebuilding the complete marker set. Thirteen reviewed collision records share true navigation anchors. Every record remains present at its original point and accessible through search, results, zoom, and selection; no presentation offset, count marker, or clustering is applied.

The calm basemap treatment is an orientation layer, not project evidence. Water and waterways are pale blue, land is warm near-white, neighboring geography is muted, and the source-backed India outline and internal orientation remain more prominent. The compact map key explains the marker channels without turning the main canvas into a permanent legend panel.

6. The Overview is a sourced public brief

The Overview summarizes the existing structured record in plain language. Release 53 makes that first read compact and type-aware: identity and location, source-backed fact chips, a 50–75 word deterministic brief, applicable facts, one meaningful change when supported, compact milestones, exact organization roles, evidence freshness, missing records, and spatial precision. Timeline, Money, Organizations, Packages, Contracts, and Evidence remain the deeper investigation views.

The displayed values, dates, roles, spatial statements, comparisons, and limitations come from the structured record and linked official evidence. Project opening and marker preview make no live AI request. If the record cannot support a conclusion, the brief leaves the field Not published or Unknown instead of filling the gap, and it omits non-applicable development-cost or construction-deadline rows from asset records.

Cost movement appears only when the two fields belong to the same compatible project record. “No upward revision reported” does not certify that a project is healthy, and a revised cost is not labeled actual or final cost. An attention signal reports why a value or date deserves review, then explicitly separates that observation from cause, blame, waste, misconduct, or corruption.

Evidence tiers describe coverage only. A Basic official record has source-backed identity and geography but may lack package or contract joins. Spatially validated, Timeline validated, Contract-linked, Payment-linked, and Deep evidence dossier describe successively different linked evidence dimensions. They do not rank whether a project or organization is good, bad, efficient, delayed, compliant, or trustworthy.

7. Breadth records pass a staged truth contract

The first breadth lane starts with 1,217 public NHAI toll-plaza identities. A deterministic SHA-256 order, a maximum of four records per state, and the minimum publish contract produce 80 reviewed records across 23 states and union territories. Release 25 adds one deterministic AMRUT Water Supply or Sewerage / Septage record for every one of its 35 listed states and union territories. Release 27 extracts all 1,987 ongoing rows from MoSPI's May 2026 PAIMANA Flash Report and begins a reviewed major-project lane. Release 38 preserves the first 75 records and extends that lane to 100; later releases grow it to 1,161 accepted records without changing the earlier order. Release 29 starts a separate IBM mine-asset lane with one strongly evidenced record, and Release 30 adds six more mine assets without weakening the mine identity or evidence contract. Release 51 adds two Digital Bharat Nidhi physical networks. Releases 68 through 86 add auditable NWIC water assets, CEA generation assets, DGCA aerodromes, major ports, National Waterway systems, Amrit Bharat station developments, completed and under-construction transmission projects, a state-balanced AMRUT batch, PNGRB pipeline assets, prioritised MMLPs, NICDC industrial-corridor projects, NHAI national-highway projects or packages, Ministry of Railways/PIB railway projects, Ministry of Jal Shakti/PIB Namami Gange sewerage projects, and PIB BharatNet State/UT systems. The public map now contains 7,609 source-backed records: 5,536 Major projects, packages, assets, or systems and 2,073 Facilities or local schemes, including 89 toll plazas behind their explicit opt-in layer.

The PAIMANA table keeps Original Cost, Anticipated Cost, Cumulative Expenditure, date of approval and start month, original or target completion, revised completion, and physical progress as separate published fields. Cumulative Expenditure is not relabeled as a contractor payment, actual cost, or final cost. The reporting agency is displayed only as the PAIMANA reporting agency, while its line ministry or department is separate; no construction contractor, concessionaire, operator, or selected bidder is inferred.

The report's final note says six project codes are omitted because their cumulative-expenditure data is inconsistent, yet two of those codes still appear in the table. The pipeline records that internal contradiction and excludes all six note-named codes through Release 66. It does not silently reconcile the report or publish the affected rows.

The AMRUT directory footer reports 8,777 projects and ₹1,97,191.85 crore in Project Cost. The displayed state rows sum to ₹1,97,191.73 crore, so the ₹0.12 crore source-display difference remains explicit. The 35-record batch is a state-balanced first sample, not a claim that all 8,777 identities have been reviewed. A malformed empty Karnataka city row is rejected before selection because it lacks a project geography identity.

Every published breadth record passes candidate, retrieved, normalized, deduplicated, entity-resolved, spatially validated, evidence-reviewed, publishable, and published stages. Stable source ids, state, coordinates, highway, chainage, supported corridor or office scope, freshness, missing evidence, per-record stage history, record hashes, and the dataset hash are committed with the snapshot. A later refresh can supersede a record without erasing the earlier source state.

The original Release 20 spatial stage checked coordinate order, India bounds, the NHAI state vocabulary, highway, chainage, corridor, and office context. Release 22 adds a separate hashed containment ledger instead of rewriting that history. Release 23 adds a second hashed evidence-review ledger for the six conflicts. A reviewed administrative state can improve discovery only when official documentary evidence agrees with the independent containment result. The original state field stays attributable, and an unresolved location never receives a guessed correction.

The network ingestion path is separate from production. The public map reads the committed reviewed snapshot, so a source outage cannot remove the directory, map markers, Project Briefs, or evidence boundaries. Contact names, phone numbers, email addresses, nearby-service contacts, and embedded document payloads are discarded before the snapshot is written.

8. Discovery filters expose evidence, not verdicts

Mapped district, evidence, freshness, precision, organization, attention, reported-progress, compatible cost-change, and compatible deadline-change filters use the same structured records as the Project Brief. Missing or incomparable inputs remain explicit. The filters do not infer full project extent, exact sites, causes, actual or final cost, blame, waste, or corruption.

A freshness filter surfaces only explicit deterministic review signals, such as a passed fee-revision date, an ended source-dated contract period, or a tender status that needs a newer check. “No freshness flag” means no such signal is present in the reviewed rule set. It is not a guarantee that every field or source is current.

The URL can preserve search, map lens, state, mapped district, other filters, selected project, longitude, latitude, and zoom. Camera values describe the reader's view only. They never replace or modify the source-backed project coordinates, geometry, spatial type, precision, or conflict record.

9. Missing data stays missing

The product does not fill gaps from inference. “Not published,” “not linked,” and “map unavailable” are different states. Contractor absence in this product does not mean no contractor exists. It means the evidence join has not been completed.

10. Organization roles are not interchangeable

NHTIS publishes one combined “Concessionaire / OMT Contractor” field, the completed-projects register uses a combined “Contractor / Concessionaire” column, and the RajmargYatra breadth record uses “Collection Agency Name.” The product preserves all three source categories. None implies that the named organization built every work, won every package, or still holds the same role after the source period. A tender for a user-fee agency does not turn an unnamed future agency into the construction contractor, concessionaire, project company, or current operator. Construction contractor, concessionaire, operator, collection agency, user-fee agency, selected bidder, special-purpose vehicle, consultant, authority, and PIU remain separate roles in the target data model.

11. Package joins and source differences

A highway package is joined to a plaza only when multiple identifiers align, such as package name, national highway, chainage range, and plaza position. A high-confidence join has a source-backed chainage range containing the plaza plus aligned corridor identifiers. A medium-confidence join has strong named-corridor and adjacency evidence but an incomplete printed boundary. Confidence describes the record linkage, not the completeness of the package dossier.

If official sources publish different chainage limits or cost values, both remain visible with their source definitions. A difference becomes a cost-overrun calculation only when the source definitions, scope, and comparison periods are demonstrably compatible.

12. What contract evidence proves

A letter of award can establish the selected bidder, date, reference number, and terms expressly accepted by the authority. It does not by itself prove the final construction contractor, subcontract structure, grant disbursement, payments, amendments, completion cost, or audit outcome.

The Pune–Solapur Package I LOA accepts a ₹299 crore grant bid payable by NHAI. The award is shared package evidence for Patas and Sardewadi, not a separate award for each plaza. The product keeps that grant separate from NHTIS project capital cost, the Independent Engineer document's total project cost, user fees, awarded construction value, expenditure, and final cost.

NHAI Addendum No. 2 changed the compatible RFP Total Project Cost from ₹922 crore to ₹1,110 crore before bid submission, an increase of ₹188 crore or 20.39%. This is labeled a bid-document revision, not a later construction overrun, actual cost, or final cost. NHTIS later publishes ₹1,371 crore as project capital cost under a different label, so it is not silently appended to the ₹922 crore to ₹1,110 crore series.

The Patas and Sardewadi NHTIS pages each repeat ₹378.70 crore as cumulative toll revenue “For Entire Project Stretch” through 31 December 2016. The product stores that as one shared project quantity. It is not attributed to either plaza, doubled to ₹757.40 crore, treated as current revenue, or mixed with project cost, grant, award, payment, expenditure, or final cost.

An expenditure register can establish one recorded payment, its date, payee, reference, purpose, and amount. It does not establish total paid-to-date or final cost. A tender status page can establish the published scope, value, dates, and stage status of that tender. It does not establish an award, payment, legal liability, recovery, or a finding of wrongdoing.

A legal disclosure can establish that a claimant advanced specified claims, the amount claimed, and the status described on the disclosure date. Allegations are not findings. A claim amount is not an award, accepted liability, payment, project cost, cost overrun, loss, or proof of misconduct. If no later source is linked, the product does not carry the historic status forward as current.

The two 16 April 2024 Delhi High Court orders establish that both NHAI and Ircon Soma petitions were withdrawn in view of settlement under Vivad Se Vishwas-II. They do not publish the arbitral award amount, settlement amount, amount paid, accepted liability, or settlement instrument. The order returning NHAI's court deposit plus interest to NHAI is a deposit disposition, not a published settlement amount or project recovery.

A company annual report can establish a dated operating-company record, project scope, concession date, and ownership as described by that company. It does not make company turnover or profit into project cost, toll revenue, an arbitration amount, or final cost. When the reported concession date has passed, the product requests a newer operating source instead of presenting the older status as current.

The 6 July 2026 NHAI listing establishes that the authority solicited a user-fee agency for the bundled Laling and Chandwad plazas, with bids due 13 July and opening scheduled 14 July. It does not establish an award, selected agency, transition date, current operator, current fee collector, tender value, payment, or final cost. The active listing is dynamic, so the exact tender ID remains visible for official portal lookup while a stable direct outcome link remains an explicit evidence gap.

Each Barsoni RFP publishes Annual Potential Collection, which is NHAI's procurement estimate for the named one-year fee-collection opportunity. It is not actual toll collected, a bidder's offer, an award amount, money paid to NHAI, project cost, or final cost. The 2024 RFP also adds toilet-block upkeep, so the 2014, 2019, and 2024 APC figures are shown as a contextual published series rather than a verdict about financial performance.

The July 2014 Barsoni expenditure row establishes one ₹55,41,415 cash performance-security refund to Sahakar Global. It does not identify the originating tender, prove a specific award date or contract value, establish total payment to the organization, or convert the payee into the construction contractor or current fee agency.

The October 2011 and February 2012 Tundla registers establish four individual authority-to-payee cheque rows totalling ₹33,63,639 across only those reviewed entries. That calculated sum is not a contract value, complete payment series, monthly amount, project cost, or actual toll collection. The repeated ₹14,58,887 amount under two different cheque references is preserved without inferring either a contractual pattern or a duplicate error.

The later Tundla RFP publishes ₹47.75 crore Annual Potential Collection as a procurement estimate. Its blank financial-bid form defines Annual Remittance in the opposite money direction, from bidder to NHAI, but supplies no bidder or amount. The RFP does not connect Cannon to the solicitation, publish an award outcome, or connect the earlier payee to AE Tollway Limited in the current combined NHTIS role.

The 2022 Raha RFP publishes ₹51.80 crore Annual Potential Collection for a one-year fee-agency opportunity and defines Annual Remittance as bidder-to-NHAI money, but its blank form supplies no bidder or amount. The later executed-contract register repeats ₹51.80 crore APC and records ₹51.90 crore accepted Annual Remittance for M B Consortium under a regular contract from 9 December 2022 to 9 December 2023. The compatible difference is ₹0.10 crore or 0.19% above APC. It is a procurement comparison, not actual toll collection, revenue, project cost, profit, expenditure, overrun, performance, or final cost.

The executed register partially resolves the RFP outcome without supplying the bid-opening result, letter of award, signed contract text, actual collection, remittance receipts, amendments, or performance. NHTIS separately publishes ₹1,245.88 crore project capital cost, ₹212.84 crore cumulative revenue for the entire project stretch through 30 October 2024, and M/s M.B. Consortium in its combined Concessionaire / OMT Contractor field. The register's M B Consortium and NHTIS's M/s M.B. Consortium remain exact source strings; no construction role, transfer, legal-entity normalization, or current status is inferred.

Raha also preserves dated source differences. The coordinate annexure and NHTIS say Nagao-Daboka-Udauli, while the RFP and executed register describe the Nagaon to Guwahati section from km 205.000 to km 273.489. The annexure names PIU Nagaon, the executed register's PIU cell visibly reads “37 Nagaon,” and the later NHTIS page names PIU Guwahati. The exact Raha name and km 254.514 point agree, but the corridor descriptions and office assignments are not silently reconciled. The NHTIS fee schedule was due for revision on 31 March 2026, so its displayed ₹120 car fee is dated evidence rather than a claimed current rate.

The Pahammawlein executed-contract register publishes ₹42.58 crore Annual Potential Collection and ₹42.81 crore accepted Annual Remittance for one regular contract with M/s Skylark Infra Engineering Pvt. Ltd. from 18 November 2022 to 18 November 2023. The compatible difference is ₹0.23 crore or 0.54% above APC. It compares a bidder-to-NHAI contract amount with the authority's procurement estimate, not actual toll collection, project cost, profit, expenditure, overrun, performance, or final cost.

NHTIS separately publishes ₹536 crore as Pahammawlein project capital cost, ₹24.41 crore as cumulative toll revenue for the entire project stretch through 31 December 2016, and MB Consortium in its combined Concessionaire / OMT Contractor field. The product does not infer a relationship, transfer, or present-day role between MB Consortium and the source-dated Skylark contract. The published contract period and fee-revision date have passed, so current agency, rate, revenue, remittance receipts, and performance need newer evidence.

The Sadahalli RFP publishes a ₹680 crore Estimated Project Cost and ₹13.60 crore bid security. It also publishes a 23 November 2009 bid due and opening schedule, but the reviewed source does not establish the actual submission date or bidder list. Its repeated Bijapur-Hungund footer conflicts with the project-specific Devanahalli-Bengaluru title, package number, chainage, and later signed agreement, so the mismatch remains visible instead of being silently corrected.

The Sadahalli signed agreement records NHAI's 8 February 2010 letter of award to selected bidder Navayuga Engineering Company Limited, the bidder-promoted project company Navayuga Devanahalli Tollway Private Limited, a ₹61.20 crore Equity Support grant, ₹34 crore performance security, 20-year concession, and 730-day scheduled upgradation period. The agreement date is 30 April 2010. That is 81 days after the LOA and 36 days beyond the stated 45-day signing window, but no reviewed extension or waiver is linked, so the interface reports the interval and evidence gap without calling it a delay, breach, fault, or misconduct.

Sadahalli's sources contain four separately scoped organization identities: the selected bidder, the original project company or concessionaire, the organization in NHAI's completed-projects combined role, and the later organization in NHTIS's combined role. Only the selected bidder's promotion of the original project company is established in the reviewed agreement. No relationship to Navyuga Road Projects or Athaang Devanahalli Tollway is inferred.

The RFP estimate, agreement Total Project Cost ceiling input, and NHTIS capital cost each publish ₹680 crore, but their definitions and dates remain separate. The ₹61.20 crore grant, ₹34 crore performance security, ₹13.60 crore bid security, and ₹458.60 crore project-stretch cumulative toll revenue through 31 December 2016 are different money types. None is labeled actual or final construction cost, and the historic toll-revenue amount is not presented as current.

13. Reviewed source ledgers

Releases 11 through 17 begin separating reviewed source facts from presentation code. The Barsoni, Tundla, Pune–Solapur, Sadahalli, Raha, and Pahammawlein trails are stored as machine-readable ledgers with entity keys, page references, document hashes, typed money fields, source differences, and outcome boundaries. Tundla validates the exact current point separately from the rounded RFP chainage and keeps payment direction distinct from a bidder's possible remittance. Pune–Solapur validates one package, two facilities, shared project money, compatible and incompatible cost definitions, and two conflicting official coordinate records. Sadahalli validates one exact plaza-to-package join while keeping procurement schedule, signed agreement timing, four organization roles, equal-number-but-different-definition cost fields, and historic toll revenue separate. Raha now validates a solicitation-to-executed-contract trail while keeping the missing bid-opening, award, signed-contract, receipt, performance, current-status, and construction records explicit. Pahammawlein retains the first reviewed executed fee-agency contract row, including a compatible APC-to-remittance comparison while keeping current status, capital cost, historic revenue, and organization roles separate. The application imports those reviewed records, and a deterministic validator checks all six ledgers before the production build. This is an early ingestion building block, not yet a nationwide automated pipeline.