docs: dissolution-inventory — post-#3338 count retcon and merge-gate alignment

[briansrls]

Auto-opened by session-dashboard for session lively-badger-465.
Pushing to session/lively-badger-465 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

• Title describes the change (not the session id or branch).
• PR body summarises what and why (replace the TODO below).
• Tests run: name the command (e.g. npm test, cargo test) and the result.
• If this closes a work item, the body contains a Closes #N directive.
• No commits on this branch are surprises (no fork/cherry-pick I did not make).
• No secrets / credentials / large binaries staged.

Summary

Reconciles docs/audit/dissolution-inventory.md to post-#3338 ground truth: P9 llvm_instruction_cost match-arm count (25 arms / 24 constructors), full test/claim roll-call in §2.8 (11 files), §2.5 workflow cores (bootstrap.dag / ci.dag fn counts and #3213-held in-file tags), TypeScript INVALID-gate and six-language scalar/bool narrative where applicable, P1 row aligned with main after merge, and hygiene (drop stale archived session id). Motivation is merge-gate / inventory accuracy so reviewers and automation can trust counts against the live tree.

Test plan

• cargo fmt --all --check — pass (docs-only change; full cargo test --workspace deferred to CI).

[briansrls]

Response to relayed cursor/composer-2 review (artifact /api/reviews/14808/artifacts/stdout.log, ~2026-05-19T11:00Z)

This entry is Verdict: APPROVE, not a request for code or doc edits. I re-checked the current tree against the review’s claims:

• Only docs/audit/dissolution-inventory.md is in scope for this work; no changes to THESIS.md, INVARIANTS.md, CODING.md, or TESTING.md.
• The inventory pins main @ e5bde4943, states 73 src/v4/**/*.dag files (not 67), and §3 frames 0 INVALID-GATE post-fix-language-files #3338 with the TypeScript D2-shaped bucket retired by fix-language-files #3338; remaining backlog is VAGUE / VALID / E-6(b) staging as described.
• §2.5 (workflow cores), §2.6 (lens + llvm_instruction_cost arm accounting), and §2.8 (test/claim roll-call, 11 files) match the review’s summary.

Action: no commit pushed (nothing to fix).

Merge readiness (GitHub-native signals): mergeStateStatus CLEAN, mergeable MERGEABLE, latest ci / affected / fmt checks pass. Formal GitHub PullRequestReview objects on this PR are empty from the API here; please rely on the dashboard’s api-review ledger for the ≥2 distinct APPROVE rule (including openai vs codex weighting) and for any REQUEST_CHANGES that do not surface as native GitHub reviews.

— sent from lively-badger-465

[briansrls]

Response to relayed codex/codex-default review (artifact /api/reviews/14814/artifacts/stdout.log, ~2026-05-19T11:11Z)

This thread item is APPROVE, not a blocking finding, so there is no fix commit.

I independently re-verified the claims codex spot-checked:

• find src/v4 -name '*.dag' → 73 files (matches the inventory).
• find src/v4/test/claim -name '*.dag' → 11 files (matches §2.8).
• src/v4/lens/cost.dag llvm_instruction_cost is a match on LlvmInstruction with 25 arms; Conversion is split into OpBitCast vs other (OpBitCast → 0, other Conversion → 1), consistent with the doc’s “24 constructors / split Conversion” accounting.

No line in the diff contradicts the cited governance docs on inspection; the change remains documentation-only.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci job SUCCESS on the latest completed run. The ≥2 distinct api-review APPROVE rule still has to be confirmed on the dashboard ledger (native GitHub /pulls/3379/reviews is empty here); together with the earlier cursor/composer-2 APPROVE artifact, that should satisfy the two-lane bar if the operator treats both as distinct api-reviews.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 6f1c11cb · Trigger: schedule
• Thinking: 194s wall

---

BLOCKING (1)

Root Cause

• docs/audit/dissolution-inventory.md workflow helper inventory is acknowledged in §2.5 but not rolled into the ranked dissolution plan/counts → enumerate the LB-P10-3213 workflow rows here or narrow the merge-gate claim so the debt remains bounded and checkable

⚠️ The retcon mostly tracks the new baseline, but the newly included workflow surface needs to be fully accounted for in the inventory before this can be treated as merge-gate aligned.

[briansrls]

Blocking review (workflow §2.5 vs §1.1 plan) — addressed in f0fd45017

The finding was valid: §2.5 counted merge-gate workflow fns but the prior text implied remaining fold ladders could sit outside the ranked queue.

Fix: Added a §1.1 “Workflow — PR #3213 merge-gate rollup” block that maps every DECISIONS.md Part 7 LB-P10-3213-* sub-row to concrete symbols (bs_member, ci_member, … ci_job_needs), names KAHN (ci_kahn_fixpoint, ci_acyclic) as 🟢 terminal per DECISIONS, routes LB-P10-3213 dissolution substrate to P2 (std/collection.dag Wave-A2), LB-P4-3213 → P4, LB-T22-3213 → T-22, and extends the P2 table cell to point at that roll-call. §2.5 now cross-references that rollup (and calls out bs_list_eq as SETEQ) instead of “triage deferred to burn-down lane.”

[briansrls]

Follow-up: origin/main had diverged in the same §2.5 region; merged 719645d75 (conflict resolved by keeping the §1.1 workflow rollup + §2.5 roll-call from f0fd45017). Branch now includes latest main (e.g. DECISIONS.md / cpp.dag delta from upstream merge c97f8a54c).

[briansrls]

Worker response (dashboard relay of prior resolution comment)

The quoted thread is not a new finding — it is the same resolution text already posted for the blocking inline review (commit f0fd45017, branch tip now 719645d75 after merging origin/main). I re-checked docs/audit/dissolution-inventory.md: the §1.1 workflow (PR #3213) merge-gate rollup is present (from ~L166), LB-P10-3213-* symbols are rolled to P2 / P4 / T-22 as described, and §2.5 no longer defers workflow list-op triage to an out-of-plan burn-down lane.

Action: no additional commit — nothing left to implement for this relayed item.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass on the latest run. Confirm ≥2 distinct api-review APPROVE artifacts on the dashboard ledger (native GitHub review list may stay empty).

— sent from lively-badger-465

[briansrls]

Worker response (dashboard relay of prior follow-up comment)

This item is the same follow-up already posted after merging origin/main: PR branch tip is 719645d75101e87056b8e56eabc6f4c0bcf196f7, which is the merge commit that brought in upstream (including the DECISIONS.md / cpp.dag delta from the c97f8a54c side of that merge) while retaining the §1.1 workflow rollup and §2.5 roll-call from f0fd45017.

Action: no commit — merge/rebase hygiene is already done; nothing further to push for this relay.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass on latest completed run. Per policy: no gh pr merge from here. Dashboard still owns the ≥2 distinct api-review APPROVE ledger.

— sent from lively-badger-465

[briansrls]

Response to relayed claude/claude-opus-4-7 review (artifact /api/reviews/14824/artifacts/stdout.log)

This entry is Verdict: APPROVE — not a change request. I re-read docs/audit/dissolution-inventory.md at branch tip against the review text: the diff remains docs-only (no hand-Rust / no EXPECTED_HAND_AUTHORED_TEST surface), §1.1 includes the workflow (#3213) merge-gate rollup binding LB-P4-3213 / LB-P10-3213 / LB-T22-3213 to P4 / P2 / T-22 with a DECISIONS.md Part 7-aligned symbol roll-call and KAHN called 🟢 terminal, and §2.5 enumerates bootstrap/CI symbols accordingly. That matches the review’s “tighter accounting” summary; I did not find a line that contradicts INVARIANTS.md §P5 for this PR’s scope (documentation-only dissolution inventory).

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, latest ci / affected / fmt pass. Confirm ≥2 distinct api-review APPROVE artifacts (including openai-pro vs codex weighting) on the dashboard ledger; no gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Response to relayed cursor/composer-2 review (artifact /api/reviews/14828/artifacts/stdout.log)

Verdict: APPROVE — treated as a positive review, not a defect list. I checked the PR’s net change vs origin/main: only docs/audit/dissolution-inventory.md differs. The inventory does what the review states: §1.1 workflow rollup binds filled src/v4/workflow cores to P2 / P4 / T-22 with DECISIONS.md LB-*-3213 mapping and a pinned-baseline fn count frame in §2.5; the prior “defer to burn-down unless flagged” posture for those folds is gone in favor of the explicit roll-call.

No rubric conflict spotted on a line-by-line read for modeling-discipline 🟡 planning, THESIS / pure-bootstrap framing, or INVARIANTS §P5 hand-Rust receipts — this PR does not add compiler authority or v3 test paths.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass on the latest completed run. ≥2 distinct api-review APPROVE rows: confirm on the dashboard ledger (cursor + claude + codex artifacts may satisfy the bar depending on provider-distinctness rules). No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 719645d7 · Trigger: schedule
• Thinking: 194s wall

---

BLOCKING (1)

Root Cause

• docs/audit/dissolution-inventory.md workflow triage split the authority/coverage obligation from the collection-combinator obligation → add the wrapper to the LB-P10 ALL/P2 rollup and update the P2/burn-down counts so the merge-gate surface is checkable.

⚠️ The workflow rollup is close, but one hand-rolled list fold remains outside the dissolution inventory.

[briansrls]

Blocking codex review (ci_all_commands_authority_ok vs P2) — fixed in 7979ae864

Verify: ci_all_commands_authority_ok is a universal fold over jobs calling ci_command_authority_ok (src/v4/workflow/ci.dag L173–177). Only ci_command_authority_ok carries the LB-T22-3213 // tag; the wrapper still shares the LB-P10-3213-ALL / all combinator dissolution class.

Doc change: §1.1 workflow rollup ALL roll-call now includes ci_all_commands_authority_ok with explicit P2 + T-22 dual binding; intro sentence covers the untagged wrapper; LB-T22-3213 bullet cross-references the ALL line; §2.5 enumerates five ci_all_* predicates including the wrapper; Caveats add item 3 so burn-down ~36 is not double-count inflated.

No DECISIONS.md edit — Part 7 LB-P10-3213-ALL table still lists the four graph predicates verbatim; the inventory states the fifth fold is the same P2 combinator pattern composed with the T-22-tagged inner check.

[briansrls]

Response to relayed cursor/composer-2 review (artifact /api/reviews/14840/artifacts/stdout.log)

Verdict: APPROVE — no defect list to fix. I spot-checked src/v4/workflow/ci.dag and bootstrap.dag at 7979ae864: CiCommand + LB-P4-3213 tag L22–23; ci_id_occurrences L79–80 + LB-P10-3213 tag; ci_command_authority_ok L162–163 + LB-T22-3213 tag; ci_all_commands_authority_ok L173–177 (universal fold over jobs); bs_member L56–57, bs_list_eq L63–71 — all align with docs/audit/dissolution-inventory.md (§1.1 workflow rollup, dual P2/T-22 on the jobs sweep, five ci_all_* call-out in §2.5, Caveat 3 against baseline double-count). Diff remains docs-only; no THESIS / pure-bootstrap narrowing and no v3 hand-Rust.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass on run 26097031883. Confirm ≥2 distinct api-review APPROVE rows on the dashboard ledger. No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Worker response (dashboard relay of prior resolution comment)

The quoted text is the same “Blocking codex review … fixed in 7979ae864” note already on the PR; branch tip 7979ae86442ba658f2190180a016a6ca00b47dbe still contains that dual P2/T-22 binding for ci_all_commands_authority_ok in docs/audit/dissolution-inventory.md.

Action: no further commit for this relay.

— sent from lively-badger-465

[briansrls]

Response to relayed codex/codex-default review (artifact /api/reviews/14843/artifacts/stdout.log)

Verdict: APPROVE — no actionable defects. I re-read docs/audit/dissolution-inventory.md at tip: the §1.1 workflow merge-gate rollup begins ~L166 (ranked-plan binding for LB-P4-3213 / LB-P10-3213 / LB-T22-3213, ALL roll-call including ci_all_commands_authority_ok dual P2/T-22); §1.2 Caveat 3 is ~L252–256 (no baseline double-count while preserving T-22); §2.5 workflow/ci.dag block ~L508–530 expands ci_all_commands_authority_ok and the five ci_all_* predicates with a pointer back to §1.1. Net effect matches the review: obligations are explicit and ranked, consistent with INVARIANTS.md §P5 as documentation of dissolution progress (no substrate or hand-Rust edits in the PR diff).

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass (run 26097031883). Confirm ≥2 distinct api-review APPROVE artifacts on the dashboard ledger. No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 7979ae86 · Trigger: schedule
• Thinking: 212s wall

---

✅ The PR cleanly resolves the prior workflow merge-gate coverage concerns.

[briansrls]

Response to relayed codex schedule review (sha 7979ae864, artifact in thread)

Verify: Branch tip 7979ae86442ba658f2190180a016a6ca00b47dbe matches the cited commit. docs/audit/dissolution-inventory.md still carries the §1.1 workflow merge-gate rollup (including ci_all_commands_authority_ok in the ALL / P2 roll-call with P2+T-22 dual binding), §2.5 alignment, and Caveat 3 guarding baseline double-count — so the “prior workflow merge-gate coverage” gap the blocking pass flagged is closed in-tree.

Action: no commit (positive closure note only).

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt pass on run 26097031883. No gh pr merge from here. ≥2 distinct api-review APPROVE rows remain dashboard-authoritative.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 00dd079f · Trigger: schedule
• Thinking: 182s wall

---

BLOCKING (1)

Root Cause

• src/v4/DECISIONS.md LB-P10-3213-ALL ledger was not updated with the newly dual-counted ci_all_commands_authority_ok obligation → add that symbol to the authoritative Part 7 row or change the inventory to point at a real receipt.

⚠️ One workflow dissolution obligation is now asserted in the inventory without matching ledger backing.

[briansrls]

Blocking codex review (ledger vs inventory for ci_all_commands_authority_ok) — addressed

Verify at 00dd079f: src/v4/DECISIONS.md Part 7 LB-P10-3213-ALL still listed only four helpers — the inventory’s P2 / ALL binding for ci_all_commands_authority_ok had no matching ledger row (valid blocking concern).

Fix (now on branch tip c65025a99):

• src/v4/DECISIONS.md: LB-P10-3213-ALL row extended to ci_all_job_ids_unique, …, ci_all_commands_authority_ok, obligation column updated to five all(...) compositions and an explicit dual LB-T22-3213 note on ci_command_authority_ok (landed in 75f6bc88a).
• docs/audit/dissolution-inventory.md: §1.1 ALL roll-call prose tightened in c65025a99 so all five symbols are cited as one DECISIONS Part 7 row (no “four + and” split).

Note: Latest commit message mentions DECISIONS but that file’s diff is in 75f6bc88a; c65025a99 is the inventory alignment commit.

CI re-running after push; wait for green affected/ci.

[briansrls]

Worker response (dashboard relay of prior resolution comment)

This thread item duplicates the “Blocking codex review … addressed” note already posted on the PR. Verify: tip c65025a993825edd14d5c238083aa552cbf65f50 still has src/v4/DECISIONS.md Part 7 LB-P10-3213-ALL listing ci_all_commands_authority_ok with the five-predicate / dual LB-T22-3213 text, and docs/audit/dissolution-inventory.md §1.1 ALL roll-call lists all five symbols under that single ledger row.

Action: no further commit for this relay.

Merge readiness (GitHub): ci, affected, fmt, and v4 pass on run 26104778542; mergeable / mergeStateStatus were UNKNOWN at poll time (GitHub mergeability recalc lag — re-check in UI). No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Follow-up: origin/main had diverged again (§1.1 P1–P4 table vs this branch). Merged dff1960a1 — conflict resolved by taking main’s P1 / P3 / P4 rows and keeping the session P2 “Workflow (#3213) … rollup” sentence in the P2 unblocks cell. CI is re-queuing on the new tip.

— sent from lively-badger-465

[briansrls]

Response to codex APPROVE_WITH_COMMENTS (artifact /api/reviews/14873)

Verify: Valid inconsistency — §1.1 P2 claimed workflow / LB-P10-3213 rolls under P2 while the 5 named + 29 sites head count stayed Section-2-only, so plan vs count read as contradictory.

Fix (8839d0cbc): §1.1 P2 🟡-count now states explicitly that 5+29 is §2.2 std/node + verilog only, and that workflow / LB-P10-3213 is a disjoint merge-gate bucket dissolved by the same P2 arrival (not folded into 5+29). §1.2 P2 lands row-local unit and dissolve text updated to list both buckets. Caveat 4 records the scoping rule for future readers.

[briansrls]

Worker response (dashboard relay of prior PR comment)

This item duplicates the codex 14873 / P2 count scoping resolution already posted: tip 8839d0cbcfafbe60ed3f8dcc66cc6591ab2d3fef still carries Caveat 4 and the scoped §1.1 P2 🟡-count / §1.2 P2 lands text from 8839d0cbc.

Action: no further commit for this relay.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt / v4 pass on run 26105365905. No gh pr merge from here. ≥2 distinct api-review APPROVE artifacts remain dashboard-authoritative.

— sent from lively-badger-465

[briansrls]

Responses to relayed api-review artifacts

1) openai-pro / gpt-5.5-pro (artifact /api/reviews/14874/artifacts/stdout.log)
The body is mostly chatgpt-reviewer runner log; the dashboard marks APPROVE. Verify: reviewed SHA prefix 8839d0cb matches tip 8839d0cbc. Net PR change remains documentation + DECISIONS.md ledger (workflow merge-gate / LB-P10-3213-ALL / P2 scoping); no new hand-Rust or compiler substrate in the diff I checked.

2) claude / claude-opus-4-7 (artifact /api/reviews/14877/artifacts/stdout.log)
Verdict: APPROVE — agreed on substance. DECISIONS.md Part 7 LB-P10-3213-ALL lists five ci_all_* predicates including ci_all_commands_authority_ok with the dual LB-T22-3213 note. docs/audit/dissolution-inventory.md §1.1 rollup + §2.5 match; Caveat 3 covers dual obligation without baseline double-count; Caveat 4 (post–codex 14873) scopes 5+29 vs workflow / LB-P10-3213 under the same P2 arrival.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt / v4 pass (run 26105365905). Together with earlier cursor/codex APPROVE artifacts, the ≥2 distinct api-review bar should be satisfiable on the dashboard ledger (openai-pro lane satisfied here). No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 8839d0cb · Trigger: schedule
• Thinking: 176s wall

---

✅ The PR now makes the workflow list-op dissolution accounting checkable against the ranked P2 plan and the Part 7 ledger, with no blocking concerns.

[briansrls]

Response to relayed codex schedule review (sha 8839d0cb, artifact in thread)

Verify: Positive closure matches tree at 8839d0cbc: §1.1 P2 scopes 5+29 to §2.2 std/node/verilog and names workflow/LB-P10-3213 as a disjoint merge-gate bucket on the same P2 arrival; §1.2 P2 lands lists both; DECISIONS Part 7 LB-P10-3213-ALL backs the five-predicate roll-call (including ci_all_commands_authority_ok) with the dual LB-T22-3213 note.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, ci / affected / fmt / v4 pass on run 26105365905. No gh pr merge from here.

— sent from lively-badger-465

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: bcb701fb · Trigger: schedule
• Thinking: 211s wall

---

✅ Design/docs-only PR; the dissolution inventory and DECISIONS ledger now align with the workflow helper surface, with no blocking concerns.

[briansrls]

Response to relayed codex schedule review (sha bcb701fb, artifact in thread)

Verify: Positive closure matches branch tip bcb701fbfe1badec4dea4bdd5134939c67ddf805 (merge-from-main tip). docs/audit/dissolution-inventory.md still ties workflow/*.dag obligations to §1.1 (P2 scoping: 5+29 vs workflow/LB-P10-3213 buckets) and §2.5 roll-call; src/v4/DECISIONS.md Part 7 LB-P10-3213-ALL still lists ci_all_commands_authority_ok with the dual LB-T22-3213 note — inventory and ledger stay aligned; no blocking discrepancy found on spot-check.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, isDraft false, ci / affected / fmt / v4 pass on run 26114267538. No gh pr merge from here. ≥2 distinct api-review APPROVE rows: confirm on dashboard ledger.

— sent from lively-badger-465

[briansrls]

Response to relayed codex/codex-default review (artifact /api/reviews/15005/artifacts/stdout.log)

Verdict: APPROVE — no blocking items to fix. I re-checked docs/audit/dissolution-inventory.md at tip bcb701fb: §1.1 P2 (~L145) keeps 5+29 scoped to §2.2 std/node/verilog and names workflow/LB-P10-3213 as a disjoint merge-gate bucket on the same P2 arrival; the workflow rollup block (~L168+) matches the review’s “single-authority / dissolution accounting” read. src/v4/DECISIONS.md Part 7 LB-P10-3213-ALL (~L1861) still lists five predicates including ci_all_commands_authority_ok with the dual LB-T22-3213 obligation — consistent with the inventory, not diluting it.

Diff remains documentation + ledger; no pinned-authority breach spotted on changed lines.

Action: no commit.

Merge readiness (GitHub): mergeable MERGEABLE, mergeStateStatus CLEAN, isDraft false, ci / affected / fmt / v4 pass (run 26114267538). No gh pr merge from here (operator merge path). ≥2 distinct api-review APPROVE rows: confirm on dashboard ledger.

— sent from lively-badger-465