Heritage Foundry End-To-End Process Timeline

Summary

The work began with repository/source access review on 2 August. The full unattended implementation ran from the first feature commit at 2026-08-03T09:02:38Z to the final release evidence uploads at about 2026-08-03T13:49:21Z: 4h 46m 43s of publicly evidenced commit-to-release activity inside a much longer research/build task.

The six original PR and Pages runs did not fail. The later cycle-free release closure deliberately failed closed four times while exercising previously untested link, genuine-Chrome and R2 contracts. Every correction changed only assurance/release controls: the candidate and deployed Site were never rebuilt.

Late-Finding Chronology

First visible timestamp Class Finding Actual invalidation
2026-08-03T04:56:01Z Corpus integrity HAR resource routes and same-year continuity could produce invalid links. producer, faithful, tiny, synthetic, site, consumer
2026-08-03T05:35:17Z Explorer routing The selected-record loader assumed every route started with dataset/. app, browser journeys, site
2026-08-03T06:16:06Z Documentation and YAML YAML-LD keyword quoting and executable Links/Narrative coverage were incomplete. templates, corpora, docs, journeys, site
2026-08-03T07:14:26Z Publication capacity The assembled Site was about 989 MB, close to the 1 GB Pages limit. site assembly and publication
2026-08-03T07:15:45Z Build contamination Ignored evaluator results were copied into the Site and changed local-only counts. site discovery, receipts, published metrics
2026-08-03T07:43:56Z CRS provenance WGS84 source geometry was incorrectly labelled EPSG:27700. records, provenance, semantic graph, search, receipts
2026-08-03T08:05:11Z Geographic scope Twenty-five Cumbria HAR rows matched only because Warwick Bridge contained Warwick. source denominator and every downstream corpus plane
2026-08-03T09:22:05Z Public boot journey The first deployed candidate did not expose Search within the 30-second action bound. Pages-root routing, app shell, public journey
2026-08-03T09:50:51Z Project-root routing Reading links, slashless Pages roots and the 404 shell assumed account-root paths. site routes, evaluator, app shell, browser tests
2026-08-03T12:02:15Z Candidate binding A stable descriptor could hide a changed executable closure, requiring plane-root binding. release roots and public verification
2026-08-03T12:35:19Z Evidence recursion Putting terminal evidence in the Site would create a self-referential hash loop. promotion envelope and release assets
2026-08-03T12:53:13Z Registry synchronization Publishing the YAML-LD registry entry required three generated projections and a new app root. registry, app static, site, browser matrix
2026-08-03T13:48:57Z Release asset naming Generic results.json/results.md basenames collided during release upload. release evidence only

Pull-Request Phases

PR Phase Files touched Line change Meaning
#67 Initial implementation 3,868 +580,091 / −188 Initial complete producer, app, evaluation and publication implementation.
#68 Late correction and completion 382 +12,842 / −9,251 Late source, provenance, topology, routing and publication corrections; 354 of 382 touched files were generated heritage outputs.
#69 Promotion and terminal publication 30 +582 / −132 Status, registry, evidence and terminal-publication promotion; only one app file changed and no app source/test changed, but the whole CI matrix reran.

GitHub Run Chronology

Start (UTC) Phase Run Wall time Result
2026-08-03T09:03:51Z PR #67 pull-request-ci 30799819042 10m 58s success, attempt 1
2026-08-03T09:15:21Z PR #67 post-merge-pages 30800609874 4m 08s success, attempt 1
2026-08-03T12:12:01Z PR #68 pull-request-ci 30812594912 10m 43s success, attempt 1
2026-08-03T12:25:01Z PR #68 post-merge-pages 30813485357 3m 40s success, attempt 1
2026-08-03T13:31:49Z PR #69 pull-request-ci 30818372899 10m 40s success, attempt 1
2026-08-03T13:42:56Z PR #69 post-merge-pages 30819232224 3m 32s success, attempt 1

R1, Terminal Assurance And R2 Closure

Start (UTC) Stage Run Wall time Result Finding or passed scope Candidate / Site rebuilt
2026-08-04T11:07:47Z candidate-release-r1 30903522306 1m 14s success The candidate tag, deterministic archive, archive attestation, complete asset set and platform immutability all passed on the first release attempt. no / no
2026-08-04T11:10:44Z terminal-assurance 30903729068 5m 49s failure Observe every rendered anchor and all three intent universes no / no
2026-08-04T11:37:24Z terminal-assurance 30905570025 8m 17s failure Observe every explicitly protected rich page in genuine Chrome no / no
2026-08-04T11:59:41Z terminal-assurance 30907144661 2m 56s success The complete link closure, genuine-Chrome protected-page observations and three-engine journey passed, but the later promotion validator exposed receipt-contract mismatches that terminal generation had not checked. no / no
2026-08-04T12:04:20Z promotion-release-r2 30907485088 1m 00s failure Validate R1-compatible R2 with trusted exact controls no / no
2026-08-04T12:13:03Z terminal-assurance 30908113699 4m 38s success Terminal generation and semantic promotion validation passed with aligned timestamps, channel and receipt size handling. no / no
2026-08-04T12:18:37Z promotion-release-r2 30908527197 1m 00s failure Enforce same-commit tags and exact envelope attestation no / no
2026-08-04T12:22:57Z terminal-assurance 30908844005 4m 14s success All 13,548 canonical URLs, 11 protected pages and the 32-action three-engine journey passed with exact receipt cross-binding. no / no
2026-08-04T12:27:41Z promotion-release-r2 30909191165 1m 15s success R1 revalidation, terminal-envelope validation, envelope attestation, ten-asset closure, publication and platform immutability all passed. no / no

The first two terminal failures exposed missing network retry semantics and a Chrome-process cleanup race. The next two promotion failures proved that a downstream semantic/release validator was not being exercised early enough. Those failures are the strongest direct evidence for the recommended shift-left microfixtures and shared contracts: the final candidate commit remained 51881ccc0ce1b77346b9cd8d4462c320bf203114 throughout all nine attempts.

Publication Gates

The 27-action candidate observation remains valid historical evidence, but the machine profile still names action 28 as the final gate while the terminal journey now has actions 29–32. Future profiles should derive stage counts and gate labels from immutable receipts rather than duplicate them manually.