docs(repo): resolve Tier-4 plan-readiness audit findings by Sawmonabo · Pull Request #124 · Sawmonabo/ai-sidekicks

Sawmonabo · 2026-05-28T08:40:52Z

Summary

NS-16 — Tier-4 plan-readiness audit of Plan-005 (Provider Driver Contract + Capabilities), Plan-006 (Session Event Taxonomy + Audit Log), and Plan-007 remainder (Local IPC + Daemon Control, Phases 4–7). Swaps the audited working copies into the corpus and lands the cross-cutting contract shapes the audit ratified. Plan-005 + Plan-006 stay approved; Plan-007 flips approved → review for the Tier-4 remainder Phases 4–7 design reopen — the Phases 1–3 partial PR sequence (#16/#17/#19) is audit-evidence preserved in §Tier 1 Partial PR Sequence + §Progress Log + §Retroactive Audit Memo.

Audit runbook: docs/operations/plan-implementation-readiness-audit-runbook.md. This is the Tier-4 pass (Tier-3 = NS-15 merged in PR #118; G6 satisfied).

Decisions ratified (D1–D7 + G3 overage)

#	Decision	Disposition
D1	Status disposition	Plan-005 + Plan-006 stay `approved`; Plan-007 flips `approved` → `review` per D2
D2	Plan-007 status flip	Tier-4 remainder Phases 4–7 reopen the design surface; Phases 1–3 partial PR sequence unaffected (preserved as audit-evidence)
D3	Spec-006 integrity protocol	ADR-018 + RFC 8785 JCS canonicalizer (UTF-16 lex-sort scalar code units) for replay-key + content-hash determinism
D4	JSON-RPC state-refusal envelope	`-32603 InternalError` envelope + reuse of `transport.unavailable` code for daemon-state-degraded refusals
D5	clipanion CLI version	`clipanion@4.0.0-rc.4` exact-RC pin (Yarn ecosystem precedent + v3 API-incompat avoidance); BL-134 tracks stable-v4 upgrade
D6	OS-keystore source	`@napi-rs/keyring v1.2.0` ratified (supersedes keytar; native binding sourced from npm preserving pnpm `side-effects-cache:false` per Plan-001 PR #3 precedent)
D7	Spec-027 daemon-side rows	Rows 2/3/7a/7b/8 integrated into `security-architecture.md` (capability boundary + audit emitter wiring)
G3 overage	Non-Tasks budget	Plan-005 2.52× + Plan-007 1.53× ratified as audit-scope-justified — 100% audit-domain content, zero redesign (runbook §Failure Modes addresses recurring-across-tiers methodology overrun, not single-tier load-bearing content additions)

G5 ephemeral-ref scrub

~100 F-XXX-Y-ZZ ID prefixes + PR #N inline refs stripped from audit-introduced plan-body text per Tier-3 Plan-003 canonical precedent (PR #118). Plan-007 baseline-preserved zones (§Tier 1 Partial PR Sequence, §Progress Log, §Retroactive Audit Memo) retain 13 PR #N refs as legitimate audit-evidence; narrative content preserved verbatim with only ephemeral prefixes scrubbed.

Anchor + path corrections (pre-commit-hook-surfaced)

The audit-introducer subagents emitted citation-shape defects the docs-anchor-check hook (lychee) caught at commit time. Root-cause fixes applied in this PR (not deferred):

Plan-005 Required-ADRs — phantom ADR-004 citation with non-existent filename slug (004-event-streams-are-private-not-shared.md; real ADR-004 = SQLite/Postgres). Replaced with ADR-017 (Shared Event-Sourcing Scope) — the ADR the body's runtime_node.* driver-event emission actually invokes (line 271). Header + Preconditions + Done-Checklist all updated consistently.
Plan-005 §CP-005-6 — broken Plan-004 filename slug (004-run-state-machine-and-events.md). Corrected to canonical 004-queue-steer-pause-resume.md.
Plan-005 §CP-005-5 — two 4-up relative-path traversals (../../../../docs/architecture/contracts/api-payload-contracts.md). Corrected to 1-up (../architecture/contracts/api-payload-contracts.md).
Plan-006 §Phase 4 narrative — Spec-015 fragment slug (#idempotency-classes not in target). Corrected to canonical #idempotency-classes-and-recovery-behavior.
Plan-006 §Phase 4 narrative — forbidden .agents/tmp/research/... link target ((.) no-op markdown link to transient research path). Sentence rewritten per AGENTS.md Surface-Forward-Then-Delete anti-pattern (the load-bearing content is already surface-forwarded into this plan + the api-payload-contracts.md §Plan-006 doc-mirror).

File manifest

File	Change
`docs/plans/005-provider-driver-contract-and-capabilities.md`	Working-copy swap: 4 Phases of Tasks audited; capability-flag + `idempotency_class` + `ProviderToolMetadata` cross-cutting shapes ratified; ADR-017 added to Required-ADRs (replacing phantom ADR-004); CP-005-5 + CP-005-6 resolutions surface-forwarded
`docs/plans/006-session-event-taxonomy-and-audit-log.md`	Working-copy swap: 4 Phases of Tasks audited; RFC 8785 JCS canonicalizer ratified; `DaemonSigningKeySource` + `OsKeystoreSealedDaemonSigningKeySource` (F-006-2-02 resolution) ratified; `retention_class` discriminator column added (F-006-3-02 Phase 3 Design B); `session_snapshots` compaction-cursor columns (F-006-4-01 Phase 4 Reading (a))
`docs/plans/007-local-ipc-and-daemon-control.md`	Working-copy swap: Phases 4–7 Tier-4 audited (`run.` / `repo.` / `artifact.` / `settings.` / `daemon.*` namespace handlers + Spec-027 row 2/3/7a/7b/8 daemon integrations); status flips `approved` → `review` per D2
`docs/architecture/contracts/api-payload-contracts.md`	Plan-007 R1/R2 state-refusal envelope + Plan-006 additive method namespace shapes (`event.readAfterCursor`, `event.readWindow`, `event.subscribe`) per CP-006-4 / CP-007-N
`docs/architecture/security-architecture.md`	Spec-027 daemon-side integration rows 2/3/7a/7b/8 (capability boundary + audit emitter wiring)
`docs/specs/006-session-event-taxonomy-and-audit-log.md`	RFC 8785 JCS canonicalizer ratification; ADR-018 reference; audit-stub retention-class additive
`docs/backlog.md`	BL-134 (clipanion stable-v4 upgrade tracking; V1.1; criterion-gated on upstream stable release)
`docs/architecture/cross-plan-dependencies.md`	§6 fold (below)

Tag on merge: plan-readiness-audit-tier-4-complete.

§6 fold (cross-plan-dependencies.md)

NS-16 :::ready → :::completed; PRs checkbox [x] tier-4.
NS-17 promoted blocked → ready per the audit-chain edge NS-16 → NS-17 (audit chain remains strictly serialized through NS-21).
Ready set after this PR: {NS-17}; blocked set: {NS-09, NS-10, NS-18..NS-21}.

Findings ledger — 97 findings, 17 critical-root-causes (5.7:1 collapse ratio)

Per-plan breakdown: Plan-005 = 49 findings (5 critical-roots) · Plan-006 = 15 findings (6 critical-roots) · Plan-007 remainder = 33 findings (7 critical-roots, minus overlap = ~17 distinct closures). Healthy finding-instance / root-cause collapse ratio (~5.7:1) — each consuming phase independently surfaced the same upstream gap from its own vantage, not over-flagging (same pattern as Tier-3 NS-15: 31 findings → 4 roots = 7.75:1).

Root-cause critical → disposition:

Critical	Disposition
RFC 8785 JCS canonicalizer unratified	D3 — ADR-018 + Spec-006 RFC 8785 ratification
JSON-RPC state-refusal envelope unratified	D4 — `-32603 InternalError` + `transport.unavailable` reuse
`clipanion` CLI version pin underspecified	D5 — exact-RC pin + BL-134
OS-keystore source library unratified	D6 — `@napi-rs/keyring v1.2.0`
Spec-027 daemon-side rows ungoverned	D7 — rows 2/3/7a/7b/8 in `security-architecture.md`
Plan-007 Tier-4 design surface contested	D2 — status flip `approved` → `review`
Plan-006 `DaemonSigningKeySource` + sealed Ed25519 source	F-006-2-02 resolution (in-plan)
Plan-006 `retention_class` discriminator column	F-006-3-02 Phase 3 Design B (in-plan)
Plan-006 `session_snapshots` compaction-cursor columns	F-006-4-01 Phase 4 Reading (a) (in-plan)
Plan-005 `InterventionType` enum co-location	CP-005-6 resolution (in-plan)
Plan-005 `CapabilityDetails` wrapper + `providerFailureDetail`	CP-005-5 resolution (in-plan)
...6 additional critical-roots dispositioned per per-plan synthesis	(full disposition matrix at `.agents/tmp/research/plan-readiness-audit/REVIEW.md` decision block)

Calibration + gates

B1–B6: B1 avg-critical/phase (Plan-005=1.25; Plan-006=1.5; Plan-007=1.75) in band; B2 total 97 (per-plan: 49/15/33 — Plan-005 over upper band, justified by 4-Phase × audit-density) · B3 Tasks:blocking (Plan-005=1.5:1; Plan-006=4.67:1) in band ✓ · B4 user-review ~60–90 min ✓ · B5 7 substantive advisor calls (range expanded vs Tier-3 baseline; Tier-4 contract-density justified) · B6 1 status-flip (Plan-007; 0–1) ✓

G1–G6: G1 skeleton additions-only ✓ · G2 all 17 root criticals dispositioned (full matrix in REVIEW.md) ✓ · G3 non-Tasks diff Plan-005 2.52×, Plan-006 1.42×, Plan-007 1.53× — Plan-005 + Plan-007 FAIL the literal <1.5× rule; ratified as audit-scope-justified per D-G3 overage (100% audit-domain content, zero redesign) · G4 all Tasks Steps → Spec coverage + Verifies-invariant ✓ · G5 0 ephemeral F-XXX-Y-ZZ / PR #N refs in audited plan bodies (Plan-007 §Tier 1 Partial PR Sequence + §Progress Log + §Retroactive Audit Memo preserve 13 PR refs as audit-evidence per baseline carryover; verified against plan-readiness-audit-tier-4-baseline tag) ✓ · G6 Tier-3 committed (NS-15 merged in PR #118) ✓

Test plan

pnpm install --frozen-lockfile provisioned the worktree; all pre-commit + commit-msg hooks green (gitleaks: no leaks; lychee: 603 links / 532 OK / 0 errors / 71 excluded; docs-corpus-checks: governance cite-walk; lint-staged: prettier; commitlint: passed).
0 ephemeral F-XXX-Y-ZZ / PR #N tags in Plan-005 + Plan-006 + Plan-007 plan-bodies (Plan-007 baseline preserved zones retain 13 PR refs as legitimate audit-evidence per Tier-3 Plan-003 precedent).
Mermaid coherence: §6 ready/blocked sets consistent with edge set; NS-16 → completed, NS-17 promoted.
Anchor + path corrections verified — git show :path | grep confirms 0 remnants of ADR-004, 004-run-state-machine, 4-up traversal, #idempotency-classes standalone, or transient .agents/tmp/ link targets in the audited Plan-005 + Plan-006 surfaces.

Codex review resolution

Round 4 — `182ac84`

Eleven findings (5×P1) resolved in 182ac84; all eleven review threads replied + resolved. Headline P1: F9 — daemon.start was modeled as an in-daemon IPC handler, which is impossible (a stopped daemon hosts no IPC server to receive the call). Removed DaemonStart* from the daemon-side surface (lifecycle schemas, IPC handlers, Tier-4 method-name table, client SDK list, renderer mutating-op gate, and the action:"start" union + shared DaemonLifecycleParams in api-payload-contracts.md); cold-boot is now the ai-sidekicks daemon start CLI process-spawn path (runtime-resolved daemon path + detached spawn + DaemonHelloAck wait, T-007r-3-4). This re-aligns Plan-007 to Spec-023:76's "spawned child process of the shell" framing — the plan had drifted from the spec, not the reverse. Other P1s: F11 (DaemonKeyStore real impl + composition-root injection at Plan-022 Tier 5 per CP-007-8, not R2 — no tier inversion), F1 (R1 registers only daemon.*+settings.*; the other five namespaces attach from their owning plans), F10 (banner read is daemon-availability-tolerant), F2 (evidence_pr: 124).

Round 5 — `34ee7f7`

Seven findings (1×P1) resolved in 34ee7f7; all seven threads replied + resolved — 45/45 threads resolved, mergeStateStatus: CLEAN. Headline P1 R5-6: the round-2 stub_signature signed only the compacted-row payload projection, leaving the denormalized scalar columns (id, session_id, sequence, occurred_at, category, type, actor) — read by SQL filters and envelope reconstruction — unbound; an at-rest scalar edit could forge an actor/type while both stub_signature and the anchor over the frozen row_hash still verified. The verifier now adds a scalar-binding check (each scalar column byte-equals its payload-projection counterpart) → a new failureMode: 'stub_scalar_mismatch', an additive-MINOR enum extension 10→11 per ADR-018 §Decision #8. Other findings: R5-1 (DaemonStatusReadResult API mirror reconciled to the canonical processState shape), R5-2 (Spec-007 amended to disambiguate daemon start as a process-spawn capability, not a daemon.start IPC handler), R5-3 (separate DaemonStopRequest/DaemonRestartRequest mirror schemas carrying idleDrainDeadlineMs), R5-4 (Plan-005 DriverClient interface enumerates exactly the 7 client-facing methods, not the stale 11), R5-5 (Plan-005 storage summary lists all four Phase-2 SQLite tables), R5-7 (R2-T6 re-scoped to the emission-side securityCritical boundary R2 owns).

Round-4 deferrals, now resolved. The two round-4 "named but not chased" divergences are closed: (a) the api-payload-contracts.md DaemonStatusReadResult mirror is reconciled to the canonical processState shape (R5-1); (b) the Plan-007 CLI §Files inventory (one-vs-three daemon-lifecycle.ts) is fixed in 8631a1d, which also reconciled the broader R3 §Files block to its authoritative Tasks — added exit-codes.ts (T-3-3) + settings.ts (T-3-6), fixed the daemon-status.ts citation (T-3-3→T-3-5), and removed the session-*/meta.ts orphans (the CLI session surface is owned by Plan-001 Phase 5; --version/--help are clipanion Cli-builder built-ins). docs/specs/007-local-ipc-and-daemon-control.md:74 is now amended (R5-2) — a real defect in the approved spec's IPC-method listing, not an intentional carry-forward.

Remaining deferrals (each owned by its own pending audit). DaemonStart survives in the capability sense (the shell boots the daemon, realized by spawn) in three Tier-8 peer artifacts revisited by their own audits: docs/specs/023-desktop-shell-and-renderer.md:76,221, docs/plans/023-desktop-shell-and-renderer.md:62, docs/operations/local-daemon-runbook.md. Amending a peer plan/spec/runbook to restate something already correct in the capability sense is out of this PR's scope.

Refs: Plan-005, Plan-006, Plan-007, Spec-006, Spec-007, ADR-017, ADR-018, BL-134

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 4.8 (1M context) noreply@anthropic.com

NS-16 — Tier 4 plan-readiness audit of Plan-005 (Provider Driver Contract + Capabilities), Plan-006 (Session Event Taxonomy + Audit Log), and Plan-007 remainder (Local IPC + Daemon Control, Phases 4–7). Swaps the audited working copies into the corpus and lands the cross-cutting contract shapes the audit ratified: - api-payload-contracts.md: Plan-007 R1/R2 envelope additions (state-refusal envelope under -32603 InternalError reuse of transport.unavailable code; daemon-state-degraded refusal path); Plan-006 additive method namespace shapes (event.readAfterCursor, event.readWindow, event.subscribe) per CP-006-4 / CP-007-N. - security-architecture.md: Spec-027 daemon-side integration rows 2/3/7a/7b/8 — capability boundary + audit emitter wiring for the daemon's namespace handlers (Plan-007 remainder). - spec-006-session-event-taxonomy-and-audit-log.md: RFC 8785 JCS canonicalizer ratified (UTF-16 lex-sort scalar code units for replay-key + content-hash determinism); ADR-018 reference; audit-stub retention-class additive. - backlog.md: BL-134 (clipanion stable-v4 upgrade tracking — criterion-gated on upstream stable release; CLI ships at clipanion@4.0.0-rc.4 exact-RC pin per Yarn-ecosystem precedent + v3 API-incompat avoidance). - cross-plan-dependencies.md section 6: NS-16 → completed; NS-17 promoted blocked → ready per the audit-chain edge NS-16 → NS-17 (the §6 audit chain remains strictly serialized through NS-21 per runbook §85–87, so NS-18..NS-21 stay blocked behind NS-17). Plan-005 stays approved (D1); Plan-006 stays approved (D1); Plan-007 flips approved → review (D2 — Tier-4 remainder Phases 4–7 reopen the design surface; Phases 1–3 partial PR sequence is audit-evidence preserved in §Tier 1 Partial PR Sequence + §Progress Log + §Notes + §Retroactive Audit Memo). G3 non-Tasks budget overages (Plan-005 2.52×, Plan-007 1.53×) ratified as audit-scope-justified — 100% audit-domain content, zero redesign (D-G3 overage). B1–B6 + G1, G2, G4, G5, G6 pass; G3 overage user-ratified per REVIEW.md decision block. G5 ephemeral-ref scrub: ~100 F-XXX-Y-ZZ ID prefixes + PR #N inline refs stripped from audit-introduced plan-body text per Tier-3 Plan-003 canonical precedent. Plan-007 baseline preserved zones (§Tier 1 Partial PR Sequence, §Progress Log, §Retroactive Audit Memo) retain 13 PR refs as legitimate audit-evidence; narrative content preserved verbatim with only ephemeral prefixes scrubbed. Plan-005 Required-ADRs corrected — phantom ADR-004 citation replaced with ADR-017 (Shared Event-Sourcing Scope) that the body's `runtime_node.*` driver-event emission actually invokes; broken Plan-004 filename slug (`004-run-state-machine-and-events.md`) corrected to canonical `004-queue-steer-pause-resume.md`; two 4-up relative-path traversals corrected to 1-up (`../architecture/contracts/...`); Plan-006 Spec-015 anchor corrected from `#idempotency-classes` to the canonical `#idempotency-classes-and-recovery-behavior`; transient `.agents/tmp/...` link target removed per AGENTS.md anti-pattern. Decisions ratified (D1–D7 + G3 overage; user-ratified 2026-05-28): - D1: Plan-005 + Plan-006 + Plan-007 design soundness — approve all - D2: Plan-007 status flip approved → review for Tier-4 remainder Phases 4–7 reopen; Phases 1–3 partial remain unaffected - D3: ADR-018 + Spec-006 RFC 8785 JCS canonicalizer ratification - D4: JSON-RPC -32603 InternalError state-refusal envelope + reuse of transport.unavailable code for daemon-state-degraded refusals - D5: clipanion@4.0.0-rc.4 exact-RC pin (BL-134 tracks stable-v4 upgrade; Yarn ecosystem precedent + v3 API-incompat avoidance) - D6: @napi-rs/keyring v1.2.0 ratified (supersedes keytar; native binding sourced from npm preserving pnpm side-effects-cache:false per Plan-001 PR #3 precedent for native-binding CI traps) - D7: Spec-027 daemon-side integration rows 2/3/7a/7b/8 integrated into security-architecture.md (capability boundary + audit emitter) Refs: Plan-005, Plan-006, Plan-007, Spec-006, ADR-017, ADR-018, BL-134 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: df52b42b8e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Resolves the 8 ACTIONABLE findings Codex surfaced on the Tier-4 plan-readiness audit (PR #124). All fixes are corpus-doc edits; no code or schema migrations land here (those ship under their owning plans at code-execution time). T1 — Plan-007 I-007-9 method-name regex updated to dotted-camelCase (permits camelCase in subsequent segments per LSP/MCP precedent); canonical regex source at api-payload-contracts.md updated and downstream label-only sweep in Plan-003 / Plan-007 / Plan-008. T2 — Plan-005 usage_quota → canonical usage_telemetry category. T3 — api-payload-contracts.md ProviderDriver interface wired: resumeSession → Promise<DriverResumeResult>, getCapabilities → Promise<GetCapabilitiesResult> per the discriminated-union shapes. T4 — sessions.daemon_signing_key (Postgres) → new local SQLite daemon_signing_keys table per ADR-004 SQLite-local-state boundary; canonical-schema doc, cross-plan-deps §6 (rows 26/42/43/135), and Plan-006 narrative sweep. T5 — anchor-before-compaction protocol per Spec-006 §Post-Compaction Integrity; failureMode enum extended 7→9 (anchor_missing_for_ compacted_range, anchor_signature_invalid) per ADR-018 §Decision #8 MINOR-additive sanction; new AC #4; Plan-006 T3.2/T3.3/T4.1/I-006-3-03 rewrites + Audit Reconciliation entry; security-architecture.md Verifier Rules add audit_stub carve-outs on chain + signature checks. T6 — Plan-005 T2.3 ProviderRegistry capability gate scoped to exclude applyIntervention (ADR-011 degraded-fallback must reach driver). T7 — Plan-005 T2.1 migration shape matches canonical local-sqlite-schema (driver_capabilities per-flag + driver_tools per-tool); T2.2 dropped findBySession; T2.4 split into per-flag + per-tool persistence; new driver_tools table added to canonical schema. T8 — Plan-006 Co-Tier Phase Ordering Precondition section + T1.4 Depends wording fix. Refs: ADR-004, ADR-011, ADR-017, ADR-018, Plan-005, Plan-006, Plan-007, Spec-006 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Sawmonabo · 2026-05-28T18:43:53Z

@codex

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e6ae9466a2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Round-2 Codex review (chatgpt-codex-connector[bot]) on PR #124 surfaced one P1 and ten P2/P3 findings. All resolved in-PR as hardened contract fixes — no V1.1 deferral. P1 — post-compaction stub authenticity (Spec-006 §Post-Compaction Integrity). The round-1 anchor-before-compaction protocol proved the ORIGINAL pre-compaction range existed but left the surviving audit-stub bytes unauthenticated. Added a per-row `stub_signature` (Ed25519 over the exact canonical bytes stored in `payload`) minted by the compactor and re-verified by the integrity verifier. Sign-exact-bytes invariant: the compactor signs the identical byte string it persists; the verifier checks `stub_signature` directly over the stored `payload` bytes (strictly stronger than re-canonicalizing — catches any byte-level edit, not only semantic — and avoids a JCS round-trip). `failureMode` enum extended 9 -> 10 (`stub_signature_invalid`, additive-MINOR per ADR-018 Decision #8). Threat model: stub-edit, cross-row replay, signature-strip, and both `retention_class`-flip directions all detected. P2/P3 (Plan-006 / Plan-007 / Plan-005): - keep stub payload: T3.2 REPLACES `payload` with the stub projection (column is NOT NULL), never nulls it — preserves the replay/renderer never-silent-omission contract. - T2.7 signing-key create() returns the PUBLIC key only; the private key is reachable solely via the signer-local read() (no private-key leak). - driver_contract_meta table (Plan-005): contractVersion hydrated from local state at cold start, not a driver round-trip. - pending_anchor_uploads retry columns (attempt_count, last_attempt_at, last_error); session_events live-row partial index. - Plan-007: idleDrainDeadlineMs uses .default(5000) alone (the trailing .optional() short-circuited the default to undefined on omit); securityCritical flag added to securityOverrides; R1 projects from the Tier-1 SupervisionHooks substrate (not the R3 consumer); session.stop/restart EXCLUDE the caller's own connection from the active-client refusal check. - self-update CLI relocated R2 -> R3-PR-a (T-007r-2-6 -> T-007r-3-6a): a CLI command cannot compile in the R2 PR before R3 scaffolds apps/cli/; the reversal honors the ratification's own R3-PR-shipping intent, recorded with rationale rather than silently overwritten. cross-plan-dependencies.md + local-sqlite-schema.md mirror the new stub_signature / retention_class / driver_contract_meta ownership. Refs: Plan-005, Plan-006, Plan-007, Spec-006, ADR-018 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Sawmonabo · 2026-05-28T20:04:21Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 91ef9aaa67

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Round-3 external review surfaced 8 P2 findings (A-H) on the Tier-4 audit diff; all resolved in-place as hardened fixes with zero deferral of in-scope work. - A (Spec-006:612): Merkle force-fire leaf basis corrected to row_hash leaves. canonical_bytes are discarded at compaction, so a verifier replaying the frozen row_hash could not reproduce a canonical_bytes-leaved root. - B (api-payload-contracts:319): method-name regex examples realigned. V1 tightens the leading segment to lowercase-only, so camelCase-root examples (textDocument.didOpen) are rejected; cite workspace.executeCommand (matches). - C (api-payload-contracts:669): split ProviderToolMetadata into an ingress shape (z.input, idempotency_class optional) and NormalizedProviderToolMetadata (z.output, required post-.default). Only the normalized shape crosses the persistence / event boundary (NOT NULL driver_tools + capability_* payloads). - D (cross-plan-deps:135): removed the reversed Plan-006 dependency clause; the dependency cycle is broken and the correct relationship stays on the next row. - E (plans/005:269): narrowed the client driver.* JSON-RPC namespace 11 -> 7. The 4 session/run lifecycle ops (createSession/resumeSession/startRun/ closeSession) are daemon-internal (orchestration-invoked), reversing the 2026-05-27 ratified 11-method design with recorded rationale. The in-daemon 10-op ProviderDriver interface is unchanged; only client exposure narrows. - F (plans/007:958): renderer daemon.restart gated on a valid DaemonHelloAck mutating-op gate (I-007-19), NOT an admin-token check; path reconciled to Spec-027's canonical ./data/admin-token (the relay-admin token, not a local-IPC authority). - G (sqlite + postgres schemas + Spec-006): anchor UNIQUE key extended to (session_id, node_id, start_sequence, end_sequence) on BOTH stores. A cadence anchor and a wider compaction-covering anchor share start_sequence and must coexist; "covering anchor exists" is a coverage query, not an exact-start match. - H (local-sqlite-schema:45): retention_class column-domain CHECK added (ALTER-addable); co-presence with stub_signature is verifier-enforced since SQLite cannot ALTER-add a table-level multi-column CHECK. Two consistency nits from the E/G ripple, plus one surfaced follow-up: - E ripple: swapped the now-daemon-internal driver.createSession method-name example (plans/007:115) for the client-facing driver.listCapabilities. - G ripple: de-staled the pending_anchor_uploads UNIQUE-key comment to the 4-column key + identical-range dedup semantics. - BL-135: surfaced the out-of-Tier-4 Plan-025 admin-token path divergence (./data/trust/relay-admin-token vs canonical ./data/admin-token) for resolution in Plan-025's own readiness audit (a Spec-027-level decision). Refs: Plan-005, Plan-006, Plan-007, Spec-006, BL-135 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Plan-007's Phase R2 detail section carried a systematic row-number inversion against the canonical Spec-027 §Required Behavior table: TLS-1.3 surfaces were cited as row 2 and the first-run key ceremony as row 8, when row 2 is "refuse-to-start without encryption", row 3 is "auto-generated strong secrets", and row 8 is "TLS 1.3 minimum". Plan-007's own coverage table (lines 374-380) and file-structure overview (lines 316-317) already cite correctly -- line 379 ("8 -- TLS 1.3 minimum") directly contradicted line 591 ("row 8 ... generates daemon master key"), which proves the defect. Likely cause: a Spec-027 row renumbering whose Phase R2 citations were never realigned. Corrected (8 references, no behavior change): - TLS 1.3 listener / TlsSurface: row 2 -> row 8 (lines 590, 602, 971; plus the 969 "8 (TLS 1.3-only)" label). - First-run key ceremony / admin-token output: row 8 -> row 3 (lines 591, 603, 970; plus 588 firstRunKeysPolicy + adminTokenPath). - nonLoopbackHost: row 3 -> row 4 (bind, per the row-4 loopback-bind default). - 969 row-2 label corrected to "refuse-to-start without encryption". tlsMode stays row 2 (the encryption-required posture row 2 governs; the TLS version floor is the distinct row 8). Plan-005/006 reference no Spec-027 rows, so the drift was Plan-007-only. Surfaced by Codex round-3 review on PR #124. Refs: Plan-007, Spec-027 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Sawmonabo · 2026-05-28T21:36:48Z

@codex

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 46eb65c10e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

f9 (architecture): daemon.start is NOT an IPC method — a stopped daemon hosts no IPC server to receive it. removed DaemonStart* from the daemon-side contract surface (daemon-lifecycle.ts schemas, ipc/handlers, Tier-4 method-name table, client SDK list, renderer mutating-op gate) and the action:"start" union + shared DaemonLifecycleParams from api-payload-contracts.md. cold-boot is the `ai-sidekicks daemon start` CLI command (runtime-resolved daemon path + detached child_process.spawn + DaemonHelloAck wait, T-007r-3-4) — this re-aligns the plan to Spec-023:76's "spawned child process of the shell" framing and reverses the prior 3-handler lifecycle design. f11: DaemonKeyStore R2 ships the interface + InMemoryDaemonKeyStore stub + first-run ceremony-against-stub only; composition-root injection + production-guard assertion + real OsKeystoreSealedDaemonKeyStore land at Plan-022 Tier 5 per CP-007-8 (no tier inversion — mirrors Plan-006 CP-006-1 PiiEncryptor boundary). f1: Phase R1 registers only daemon.*+settings.*; the other five Tier-4 namespaces attach from their owning plans per NS-26 (R1 test asserts the 2 R1-owned namespaces + the I-007-6 duplicate-rejection mechanism, not all seven). f3: nested clipanion command paths (daemon start/stop/restart/status); add daemon status. f10: banner is daemon-availability-tolerant (daemon start surfaces the freshly-spawned daemon's banner after DaemonHelloAck; an unreachable daemon degrades to a typed "daemon unavailable" rather than failing the command). f2: evidence_pr TBD → 124. f4: DaemonSigningKeySource.generate() → .create(sessionId). f5: split Plan-001 forward-declared integrity columns from Plan-006-owned additive Tier-4 migrations per CP-006-6. f6: canonicalize ^3.0.0 → exact pin 3.0.0 (canonical bytes feed every event hash/signature; T2.3 golden vectors bind output to RFC 8785 §A). f7: driver.* client-facing SDK surface is the 6 non-lifecycle verbs; the 4 session/run lifecycle ops get no client-facing schema (daemon-internal). f8: security-architecture schema-migration is two-wave (Plan-001 forward-declares 4 integrity columns; Plan-006 ships the additive Tier-4 columns + daemon_signing_keys + pending_anchor_uploads tables). Refs: Plan-005, Plan-006, Plan-007

Sawmonabo · 2026-05-28T22:55:24Z

@codex

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 182ac84b01

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

reconcile the api-payload contract mirror and the approved specs to the audited plans (the contract/spec surface these plans directly implement), and close a P1 post-compaction audit-log forgeability gap. no improper deferral: the mirror and specs are fixed at the root, not punted to their own audits. - R5-1 (P2): api-payload DaemonStatusReadResult reconciled to the plan-007 canonical (processState enum + protocolVersion + transportEndpoint + uptimeMs) - R5-2 (P2): spec-007 disambiguates DaemonStart as the supervisor process-spawn capability (CLI/shell), not an IPC method -- coherent with its own Default Behavior / Example Flows; clarifying-amendment note added (status stays approved) - R5-3 (P2): api-payload DaemonLifecycle split into separate DaemonStop / DaemonRestart request schemas carrying idleDrainDeadlineMs (default 5000) + {accepted:true}; refusal stays the lifecycle_conflict error envelope - R5-4 (P2): plan-005 DriverClient enumerates the 6 non-lifecycle verbs + subscribeEvents (drops the stale "+ 8 more" implying the 11-method client) - R5-5 (P3): plan-005 storage summary lists all four Phase-2 tables (+ driver_tools, driver_contract_meta); brittle line ref -> section anchor - R5-6 (P1): post-compaction scalar-binding. the round-2 stub_signature signs only the payload projection; the surviving scalar columns (category/type/actor/ occurred_at, id/session_id/sequence) were unbound -- an at-rest scalar edit forged a filter/reconstruction value while stub_signature + anchor still verified. the verifier now binds each scalar column to the signed projection; new stub_scalar_mismatch failureMode (additive-MINOR, enum 10->11) swept across spec-006, plan-006 (T4.1 / I-006-3-03 / lines 55,228), security-architecture rule 4, cross-plan-dependencies, local-sqlite-schema - R5-7 (P2): plan-007 R2-T6 re-scoped to the emission-side securityCritical projection (R2's deliverable); the CLI --no-banner suppression test stays at R3-T8 (R2 owns no apps/cli/ file) Refs: Plan-005, Plan-006, Plan-007, Spec-006, Spec-007

The R3 §Files inventory had drifted from the authoritative Tasks (T-007r-3-1..6a) and §PR Boundary Mapping. Reconcile every row to its real task, in task-ascending order: - add `exit-codes.ts` (T-007r-3-3) — already in the §PR Boundary CREATE list (line 626) and Acceptance Criteria but missing from the inventory. - add `settings.ts` (T-007r-3-6) — `settings effective-read`; likewise one of the §PR Boundary "6 commands" but absent from the inventory. - fix the `daemon-status.ts` citation T-007r-3-3 -> T-007r-3-5 (T-3-3 is exit-codes.ts; daemon-status is T-3-5) and drop the inaccurate "NDJSON streaming" note — T-3-5 renders a single text-or-`--json` read. - remove the `session-*` quartet and `meta.ts` orphan rows: no R3 task creates them, and both are absent from the §PR Boundary Mapping (6 commands) and Acceptance Criteria. The CLI `ai-sidekicks session` surface is owned by Plan-001 Phase 5 (Spec-001:147,156), not Plan-007 R3; `--version`/`--help`/global flags are clipanion `Cli`-builder built-ins in main.ts (T-007r-3-2), not a separate command file. Self-audit follow-on to the F9 daemon-lifecycle file-count fix (the named PR-body divergence (b)); no Codex thread. Refs: Plan-007, Spec-007 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Sawmonabo · 2026-05-29T00:26:23Z

@codex

chatgpt-codex-connector · 2026-05-29T00:30:46Z

Codex Review: Didn't find any major issues. Already looking forward to the next diff.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector Bot reviewed May 28, 2026

View reviewed changes

Sawmonabo and others added 2 commits May 28, 2026 17:24

chatgpt-codex-connector Bot reviewed May 28, 2026

View reviewed changes

Sawmonabo and others added 2 commits May 28, 2026 19:34

Sawmonabo merged commit cb3d1b2 into develop May 29, 2026

Sawmonabo mentioned this pull request May 29, 2026

chore(repo): housekeeping for PR #124 — NS-16 fold + Plan-007 flip #125

Merged

chatgpt-codex-connector Bot mentioned this pull request May 29, 2026

docs(repo): promote Plan-007 to approved + Tier-4 R3 deps + census #126

Merged

Conversation

Sawmonabo commented May 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Decisions ratified (D1–D7 + G3 overage)

G5 ephemeral-ref scrub

Anchor + path corrections (pre-commit-hook-surfaced)

File manifest

§6 fold (cross-plan-dependencies.md)

Findings ledger — 97 findings, 17 critical-root-causes (5.7:1 collapse ratio)

Calibration + gates

Test plan

Codex review resolution

Round 4 — 182ac84

Round 5 — 34ee7f7

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Sawmonabo commented May 28, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Sawmonabo commented May 28, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Sawmonabo commented May 28, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Sawmonabo commented May 28, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Sawmonabo commented May 28, 2026 •

edited

Loading

Round 4 — `182ac84`

Round 5 — `34ee7f7`