S238 wrap → S239 dispatch — synthesis of pre-session audits

S238 wrap → S239 dispatch — synthesis of pre-session audits

Date: 14/05/2026 (S238 wrap-up — single canonical entry point for S239 Wave 1 dispatch) Author: S238 main session (foreground synthesis pass after 3 parallel pre-S239 audits returned) Branch head at synthesis: f0d82574 + 3 fix commits (this doc lands as part of the S238 wrap commit) Pilot reference: docs/plans/phase-0-investigation/architecture/01-vision.md (S238 commits 4b9dccd3 + 6c30640c + S238-wrap 3-tier heritage taxonomy patch)

Purpose

Three parallel sub-agents ran during S238 wrap to prepare S239 Wave 1 dispatch on the remaining 7 WP4 architecture sub-docs. Each agent ran in an isolated worktree with no visibility into the others’ work. This doc synthesises their outputs, identifies contradictions, records resolutions ratified by Liam in S238, and consolidates the S239 starting state — so S239 reads ONE doc to know where to start, not three audits + a pilot + a stack of central-register docs cold.

The three pre-S239 audits

Audit	Path	Lines	Verdict	Commit
(1) DG + core-doc consistency audit	docs/specs/core-docs-pathway-assessment/S238-pre-S239-consistency-audit.md	504	9 CONSISTENT / 1 DRIFT-DETECTED (MEDIUM) / 1 LAG-ACCEPTABLE; ZERO HIGH-severity blockers for S239 Wave 1	164937d2
(2) Architecture sub-doc construction guide	docs/specs/core-docs-pathway-assessment/architecture-sub-doc-construction-guide.md	546	10 sections; pilot-pattern template + citation discipline + no-fabrication rule + forward-ref policy + verifier gate	0b967d13
(3) Per-sub-doc readiness audit	docs/specs/core-docs-pathway-assessment/architecture-sub-doc-readiness-audit.md	524	71 sections audited; 54 READY-TO-DRAFT (76%) / 2 product-spec / 4 tech-spec / 0 investigation / 11 ratification (8 of 11 concentrate on 06-mcp-tooling Theme F)	f0d82574

Total: 1,574 lines of audit output. This synthesis is the executive layer; the three audits are the substrate.

Cross-doc cross-checks (synthesis pass)

The foreground synthesis cross-checked the three audits against each other and against the pilot. Specific cross-checks:

Cross-check	Verdict
Construction guide §9 per-sub-doc adaptations vs Readiness audit section counts	CONSISTENT — 04-workspace-types.md confirmed biggest (13 sections); 03-tech-stack.md ~9 sections; 07-collapse-list.md thin (4 sections); 06-mcp-tooling.md gated (8 sections all (d)) — both audits agree
Construction guide §3 citation patterns vs Consistency audit Q-OQR1 cross-checks	CONSISTENT — all 5 patterns in the guide resolve to verified-consistent register entries
Construction guide §6 forward-reference policy vs Readiness audit Risk 1 (04 §4 satellites)	CONSISTENT — both flag forward-references as the load-bearing discipline; readiness audit identifies 04-workspace-types.md §4 as the place forward-refs could leak
Readiness audit Wave 1 plan (03 + 07 parallel) vs INV §6 wave structure	CONSISTENT — both confirm Wave 1 = 03-tech-stack + 07-collapse-list parallel after pilot landed S238
Construction guide §10 verifier gate vs CLAUDE.md “Verification Gates” workflow	CONSISTENT — verifier-after-each-wave matches existing project discipline
Pilot §7.1 heritage-doc tags vs Construction guide §4.1 three-tier taxonomy	CONTRADICTION DETECTED — see Contradictions §1 below

Contradictions detected + resolutions

Contradiction 1 — Heritage-doc status taxonomy (RESOLVED in S238 wrap)

Found in:

• Pilot 01-vision.md §7.1: used [CURRENT-CANONICAL] for all 4 heritage docs with prose qualifiers (“Pre-Phase-0.9 in places”, “outdated in places”)
• Construction guide §4.1: introduced three-tier vocabulary [CURRENT-CANONICAL] / [PARTIALLY-SUPERSEDED] / [FULLY-SUPERSEDED]
• Readiness audit Heritage-doc disposition table: used prose qualifiers, no explicit tags

Resolution (Liam ratification S238 wrap): Three-tier taxonomy is canonical. Construction guide §4.1 is the source-of-truth.

Applied fix (S238 wrap commit):

• Pilot 01-vision.md §7.1 patched to use three-tier vocabulary:
  • Platform Overview → [PARTIALLY-SUPERSEDED] (vision framing + Five Guiding Principles hold; bid_workspaces framing stale per Q-OQR1-02)
  • Claude Integration Guide → [CURRENT-CANONICAL]
  • AI Integration Strategy → [PARTIALLY-SUPERSEDED] (§1 holds; §17+§18 build status + effort estimates pre-Phase-0.9 + out of scope)
  • Product Differentiation Audit → [PARTIALLY-SUPERSEDED] (competitive positioning holds; “AI features” framing has shifted post-S237)
• A status-taxonomy preamble paragraph added to pilot §7.1 referencing the construction guide §4.1 canonical definition.

Downstream: All subsequent sub-doc agents in S239 must use the three-tier taxonomy per construction guide §4.1.

Contradiction 2 — None substantive

No other contradictions detected. All three audits are mutually consistent on:

• Wave 1 dispatch shape (03-tech-stack.md + 07-collapse-list.md parallel)
• The lone Wave 1 pre-dispatch ratification (07-collapse-list.md thin-pointer-vs-self-contained shape)
• Theme F status (STILL-OPEN — 06-mcp-tooling.md remains gated)
• Per-sub-doc readiness verdicts
• Heritage-doc citation discipline (cite sparingly; only where heritage framing is uniquely load-bearing)
• The no-fabrication discipline (gap-flag + categorise + suggest prerequisite work)

Documentation-hygiene fixes applied (S238 wrap)

Two non-blocking drift findings from the consistency audit were fixed during S238 wrap synthesis (not deferred to S239) so S239 starts clean:

Fix 1 — feedback-findings-review.md §5.2.1 rows 6 + 7 (MED-1 from consistency audit)

Pre-fix state: Rows 6 (OQ-Q24-B audit_log RLS pattern) and 7 (OQ-Q24-C op_id propagation pattern) marked STILL-OPEN. Canonical synthesis-v2 §3.16 + §5.1 + decision-graph §11.4.1 have both as RESOLVED-S236 (WP8 ruling). The §6 closures appended at S237 didn’t retroactively update §5.2.1.

Applied fix: Both rows updated to RESOLVED-S236 with explicit “Superseded — see 00-synthesis-v2.md §3.16 + §5.1 + decision-graph §11.4.1 + this doc §6.4 for the audit-trail” cross-link.

Fix 2 — INV-architecture-split-readiness.md §3 row 1 (LOW-1 from consistency audit)

Pre-fix state: Row 1 cited Platform Overview + Claude Integration Guide at docs/client-documentation-base/markdown/... (wrong path; the -base/markdown segment doesn’t exist).

Applied fix: Path corrected to docs/client-documentation/... (matches the pilot’s §7.1 citation + the actual repo layout). Status column also updated to note the S238 landing as 01-vision.md (renamed from 01-vision-mission.md per Liam S238 direction).

What remains LAG-ACCEPTABLE (tracked, not actionable now)

LAG-1: docs/reference/SCHEMA-QUICK-REFERENCE.md Last verified 07/05/2026 + pre-Q-OQR1-16 schema state. Tracked as “apply tracks the gap” per the migration draft’s own embedded checklist (supabase/migrations/20260514150238_*.sql lines 141-152). Schema doc bumps when migrations apply; current draft is APPLY GATED ON LIAM REVIEW.

S239 Wave 1 dispatch — consolidated state

Branch head at S239 start: content-items-investigation at the S238 wrap commit (includes all 3 audits + 3 hygiene fixes + this synthesis doc).

Wave 1 targets: 03-tech-stack.md + 07-collapse-list.md (parallel worktree sub-agents).

Pre-dispatch tasks (S239 kickoff — Wave 0)

#	Task	Source	Effort	Wave-blocking?
1	Liam ratification — 07-collapse-list.md shape: thin-pointer-to-0.9-collapse-candidates.md vs fully self-contained	Readiness audit Critical-path #2; INV §6 S238 in-session decision suggestion	1-line ratification	YES (Wave 1)
2	Brief sub-agents per construction guide + readiness audit per-sub-doc spec	Construction guide §10 verifier-gate convention; readiness audit per-sub-doc deep-dive	foreground prep	YES (Wave 1)

Pre-Wave-2 tasks (defer to S239 Wave 2 kickoff, not Wave 1)

#	Task	Source	Wave-blocking?
3	Liam ratification — RLS-PATTERN destination decision (separate docs/specs/rls-pattern/{PRODUCT.md,TECH.md} vs inline in 02-data-flow.md §5 + 04-workspace-types.md §12)	Readiness audit Critical-path #1; INV §7.4 recommends separate doc	Wave 2 (NOT Wave 1)
4	Liam disambiguation — 04-workspace-types.md §4 per-application-type satellites scope (procurement-only-with-seats vs full 5-satellite)	Readiness audit Critical-path #4; Risk 1	Wave 2 main
5	Liam ratification — 08-new-features.md §1 Knowledge Map naming + §8 admin UI v1.1 mention-vs-PRODUCT.md scope	Readiness audit Critical-path #5	Wave 2 tail

Independent-track task (any wave when Theme F lands)

#	Task	Source	Wave-blocking?
6	Liam pre-decision — Theme F MCP-action review pass (mempalace direct vs wrapped in KH MCP)	Consistency audit §8 + readiness audit Critical-path #3 + Risk 3	Blocks 06-mcp-tooling.md only; orthogonal to Wave 1 / Wave 2 / Wave 3 dispatch

Refined wave sequence for S239+ (consolidated from INV §6 + readiness audit)

S239 — Wave 1 (Layer 1): 03-tech-stack + 07-collapse-list parallel
                          ↓ (after both land + verifiers gate)
S239 — Wave 2 main:       04-workspace-types foreground (BIG sub-doc)
                          ↓ (after lands + verifier gate)
S239 — Wave 2 split:      02-data-flow + 05-qa-flow parallel
                          ↓
S239/S240 boundary:       08-new-features foreground
                          ↓
S240 — Wave 3:            09-diagrams foreground
                          ↓
Independent track:        06-mcp-tooling parallel sub-agent (whenever Theme F closes)

S239 sub-agent dispatch discipline (mandatory)

Every Wave 1 sub-agent prompt MUST cite:

1. Construction guide (docs/specs/core-docs-pathway-assessment/architecture-sub-doc-construction-guide.md) — agent reads §1-§10 as standing instructions. Especially §5 (no-fabrication rule) and §6 (forward-reference policy).
2. Readiness audit (docs/specs/core-docs-pathway-assessment/architecture-sub-doc-readiness-audit.md) — agent reads its sub-doc’s deep-dive section + the cross-reference ratification anchors table.
3. Pilot (docs/plans/phase-0-investigation/architecture/01-vision.md) — agent reads as concrete pattern reference.
4. Central register — 00-synthesis-v2.md §3-§5 + 0.9-decision-graph.md §11 + INV-architecture-split-readiness.md §3-§5 (load-bearing ratifications).
5. First-action mandate — git reset --hard origin/content-items-investigation per CLAUDE.md “Worktree agents start stale”.

Verifier sub-agent dispatched after each Wave 1 sub-doc lands. Verifier reads construction guide §10 verification-gate checklist + the just-landed sub-doc + the cited ratification sources. Verdict: PASS / PASS-WITH-NOTES / FAIL with concrete fix-agent dispatch criteria.

Open carry-forwards from S238 (not blocking Wave 1)

1. WP5(b) migration draft at supabase/migrations/20260514150238_enable_rls_auto_event_trigger_and_grants_pattern.sql — APPLY GATED ON LIAM REVIEW. Combined per 00-synthesis-v2.md §3.16 + §3.17 (auto-RLS event trigger + grants pattern helper). May 30 platform deadline.
2. Theme F MCP-action review — STILL-OPEN; surface for Liam pre-decision at S239 kickoff or any later point. Blocks 06-mcp-tooling.md only.
3. Pilot 01-vision.md content review by Liam — out-of-session per S238 wrap direction. Adjustments fold into the next sub-doc dispatch wave.
4. 14 pre-existing locked agent worktrees (S237 + older) preserved per S238 direction — parallel CI-fix session + production-readiness track may use these.

Build status at S238 wrap

• bun run test __tests__/api/auth.test.ts — 11/11 pass in 760ms (was 5s flake pre-S238)
• bun run test __tests__/docs/reference-doc-edit-coupled-freshness.test.ts — 19/19 pass (post-skip-marker fix)
• Full suite run mid-session: 12,458 pass / 24 skip / 1 fail (doc-freshness — fixed via skip-marker per CLAUDE.md “Guard tests break on structural changes” gotcha)
• bun lint — clean (Babel 500KB note on lib/mcp/app-bundles.ts informational only)
• 7 commits pushed to origin/content-items-investigation during S238 main work; 4 additional commits during S238 wrap (3 audits + 3 hygiene fixes + this synthesis)

Confidence

HIGH for synthesis verdict: Wave 1 dispatch is unblocked; the lone resolved contradiction has been patched; the two MEDIUM+LOW hygiene fixes are in; the three pre-S239 audit docs + this synthesis are the canonical S239 starting state.

MEDIUM confidence on judgement-call items: heritage-doc taxonomy ratification (three-tier) is Liam’s call applied here; if Liam revises in S239 review of the pilot, the construction guide + pilot §7.1 will need a small follow-up patch. The “READY-TO-DRAFT” verdicts in the readiness audit are conservative — actual sub-doc drafting may surface additional gaps (per construction guide §5 no-fabrication discipline, those gaps get flagged in-doc, not papered over).

---

End of S238 wrap synthesis. S239 entry point: read this doc + the three audits referenced in §“The three pre-S239 audits”, then start Wave 1 dispatch per §“S239 Wave 1 dispatch — consolidated state”.