069ceb55c5
The keel names this repository as its evidence source and cites these paths by
name in its most-read documents. They existed only in a working tree.
coreai/docs/atlas/00-atlas-overview.md "Evidence base: ... the full ADR
audit (../core/docs/adr-audit/:
all 314 ADRs, 431 findings, four
governance patterns)"
coreai/docs/plans/2026-07-29-specification-reconciliation.md
"Predecessor evidence:
../core/docs/adr-audit/ ...,
the 2026-07-28 census"
coreai/docs/Whitepaper.md "Sources: ../core/docs/adr-audit/
finding AA-250 and Pattern C-bis"
A citation to a path nobody else can reach is testimony about this repository,
not evidence from it — the keel's own transit rule, applied here. Until this
commit the atlas's entire evidence base, the specification reconciliation's
predecessor evidence, and a Whitepaper source were unreachable to every reader
but one.
docs/adr-audit/ 56 files — the ADR census, stack taxonomy, per-ADR cards,
component cards, triage log, finding register, and the
cross-batch synthesis the keel's protocol step 2b reads
docs/census/ 10 files — the 2026-07-28 module census (666 modules,
293 serving-reachable) the atlas quotes
docs/plans/2026-07-29-adr-audit-plan.md
docs/adr/INDEX-by-domain.md regenerated
evals/ two obligation fixtures updated
This is the same defect the keel spent today finding in its own records, in the
other direction: there, a pin resolved against a working tree and let an
unpublished file discharge an accepted claim. Here, published prose cited
unpublished evidence. Both are the gap between what one machine can see and what
the record can prove, and both close the same way — publish, then cite.
[Verification]: uv run core test --suite smoke -q -> 786 passed (run before this
commit; docs plus two eval fixtures, no code path touched — re-run recorded on
the PR).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qw8vDY2bpYHUQYQp5SFKEG
|
||
|---|---|---|
| .. | ||
| 10-stack-dossiers | ||
| 11-adr-cards | ||
| 00-scope-and-method.md | ||
| 01-adr-census.md | ||
| 02-stack-taxonomy.md | ||
| 12-triage-log.md | ||
| 20-finding-register.md | ||
| 21-drift-report.md | ||
| 22-consolidation-report.md | ||
| 30-alignment-matrix.md | ||
| 40-triage-queue.md | ||
| 50-cross-batch-synthesis.md | ||
| 99-orchestration-log.md | ||
| MANIFEST.md | ||
| README.md | ||
| TEMPLATE-adr-card.md | ||
| TEMPLATE-stack-dossier.md | ||
docs/adr-audit/ — The ADR Audit (started 2026-07-29)
A read-only, evidence-bearing audit of every ADR in docs/adr/ against the codebase, the three engineering pillars, the seven axioms, the Whitepaper/Yellowpaper, and each other — plus a necessity/generality pass asking which components are irreducible and which duplicate what the geometric substrate already provides. Full rationale: docs/plans/2026-07-29-adr-audit-plan.md. Working charter: 00-scope-and-method.md.
If you are resuming this work cold (new session, new subagent, compacted context): read MANIFEST.md first — it's the single source of truth for what batch/phase is done vs. pending. Then 00-scope-and-method.md for the charter. Then proceed.
| File / dir | Phase | What it is |
|---|---|---|
00-scope-and-method.md |
— | The charter: pillars, axioms, discipline, tiering, execution order |
MANIFEST.md |
— | Batch tracker. Check this before doing anything else. |
TEMPLATE-adr-card.md |
— | Per-ADR fill-in template (Tier B/C) |
TEMPLATE-stack-dossier.md |
— | Per-stack fill-in template (Tier A) |
01-adr-census.md |
1 | Full 314-file inventory, numbering-collision disambiguation, phased-family list |
02-stack-taxonomy.md |
2 | ADRs mapped onto docs/assessment/02-layer-taxonomy.md's zones, tiered |
10-stack-dossiers/ |
3 | Tier A dossiers. A1–A5 = Batch 1, A2.1–A2.6 = Batch 2, Batch3–6-*-redo.md = Batches 3–6. *-consolidated.md files are RETRACTED — do not cite (see 20-finding-register.md) |
11-adr-cards/ |
3 | Tier B cards. B1–B6 = Batch 1, B2.1–B2.7 = Batch 2, Batch3-TierB-redo.md = Batch 3. *-consolidated.md = RETRACTED |
12-triage-log.md |
3 | Tier C: rapid-triage log, one row per singleton/scope-note |
20-finding-register.md |
4 | The AA-N finding ledger |
21-drift-report.md |
4 | Whitepaper/Yellowpaper/prior-ADR contradictions |
22-consolidation-report.md |
4 | Necessity/generality clusters — the parsimony pass |
30-alignment-matrix.md |
4 | Master table, one row per ADR (Batch 1 complete; Batches 2–6 verdicts live in their dossiers) |
40-triage-queue.md |
5 | Ranked 🔴 Block / 🟡 Repair / 🔵 Consolidate / 🟢 Monitor |
50-cross-batch-synthesis.md |
6 | Start here for the corpus-level result — four corpus-wide governance patterns, what the audit got wrong about itself, and the ordered remediation program |
99-orchestration-log.md |
— | Log written by the retracted external pass; retained as a record of that episode, not as audit evidence |
Status (2026-07-29): COMPLETE — all six batches, 314/314 ADRs. Batches 3–6 were redone after an external pass over them was retracted for unreliability (fabricated artifact citations, a silent severity downgrade, and near-zero critical findings in territory already proven to have them) — full evidence in 20-finding-register.md's retraction notice; AA-331–AA-438 are void and real findings resume at AA-439.
Read in this order: 50-cross-batch-synthesis.md (the result) → 40-triage-queue.md (what to do) → 20-finding-register.md (every finding) → the dossiers/cards (the evidence). MANIFEST.md is the resume point for any unfinished work.