docs(specs+board): post-#360 substrate-recognition sweep — 3 spec re-scopes + cross-source matrix + sprint prompt template + board hygiene

[AdaWorldAPI]

Summary

Post-PR #360 review pass surfaced 3 architectural defects in sprint-3 specs that recurred the very anti-pattern PR #358 named ("Designing What's Already Built"). Specs PR-A-1, PR-C-1, PR-D-1 over-scoped because they didn't sweep post-#355 substrate.

What this PR does

1. Re-scope 3 over-scoped specs (append-only corrections)

Spec	Defect	Fix	Effort delta
PR-A-1	Proposed NEW SpoQuad type; ignored shipped SchemaPtr.ontology_context_id (namespace.rs) + NamespaceRegistry::seed_defaults codebook (namespace_registry.rs)	Reuse existing types; SPO writer awareness of g + Lance schema column only	~300 LOC / 2 days → ~150 LOC / 1 day
PR-C-1	Proposed NEW GenericBridge in callcenter; ignored shipped BridgeFromRegistry trait + 3 impls (bridges/{woa,medcare,ogit}_bridge.rs) + 2 merged consumer scaffolds (woa-rs#2 ~150 LOC, medcare-rs#110 ~250 LOC)	Move trait to contract; ConsumerPointer is data on existing enumerate(namespace) projection; for_g(g) factory dispatches to existing bridges by namespace lookup	~200 LOC / 1-2 days → ~80 LOC / ½ day
PR-D-1	Proposed 600-LOC FMA-specific OWL hydrator; ignored shipped parse_ttl_directory_with_provenance + OntologyRegistry::hydrate_once_sync + attach_provenance + MappingRow.attribute_sources	Build OWL/RDF-XML adapter that emits MappingProposal + ProvenanceBundle and calls existing hydrate_once_sync. Mark Pattern D PARTIALLY SHIPPED.	~600 LOC / 3-4 days → ~250 LOC / 1-2 days

Critical-path compression: sprint-3 ~6 working days parallelized → ~3-4 days post-substrate-sweep. The genuinely-new ~5-pattern set (B/E/F/G/J) accounts for the majority of remaining design-phase work.

2. Tier-0 reclassifications

.claude/knowledge/tier-0-pattern-recognition.md updated with substrate-recognition sweep section:

• Pattern A: DESIGN PHASE → PARTIALLY SHIPPED (SchemaPtr.ontology_context_id + NamespaceRegistry codebook ship)
• Pattern C: DESIGN PHASE → PARTIALLY SHIPPED (BridgeFromRegistry trait + 3 impls + 2 consumer scaffolds ship)
• Pattern D: DESIGN PHASE → PARTIALLY SHIPPED (TTL hydrator + provenance + attribute_sources ship; OWL/RDF-XML adapter is the only net-new piece)

3. NEW: Cross-source pattern matrix

.claude/knowledge/pattern-recognition-cross-source.md — 4-taxonomy matrix mapping each architectural insight across:

• A-O Patterns (sprint-2/3 specs)
• Pillars 0-4 (palantir-parity-cascade)
• .grok/ parallel-session epiphanies (22 entries already exist; never cross-referenced)
• Shipped substrate (file:symbol citations for ~28 crates)

Plus a "shipped consumer scaffolds" section listing woa-rs#2 / medcare-rs#110 / smb-office-rs#29 / medcare-rs#98 as the 4 real precedents that already validate Pattern C's LOC reduction claim. The W8 hypothetical hubspo-rs target re-targets to "verify woa-rs + medcare-bridge hit budget" (those are merged precedents).

4. NEW: CCA2A sprint prompt template

.claude/knowledge/cca2a-sprint-prompt-template.md — formalizes the lessons learned across sprints 2 and 3:

• Substrate-grep checklist (mandatory before any new spec)
• Wrong-repo guardrail snippet (prevents the W7-sprint-2 / W9-sprint-3 recurrence)
• Pattern-letter discipline (canonical letters embedded in every prompt)
• Race-with-merge mitigation protocol
• Sprint scaffolding pattern (branch + SPRINT_LOG + worker prompts pre-created)

5. Board hygiene — sprint-3 entries

W8 (INTEGRATION_PLANS) + W9 (LATEST_STATE) sprint-2 entries are on main; sprint-3 entries were missing. This PR appends them.

What this PR does NOT change

• Pure documentation / spec corrections
• Append-only on all 5 board / spec files
• No code touched
• Original PR-A-1 / PR-C-1 / PR-D-1 spec content preserved verbatim; corrections are appended sections at end

Files changed

File	Type	Change
.claude/specs/pr-a-1-spo-g-u32-slot.md	append	post-#360 substrate sweep correction
.claude/specs/pr-c-1-generic-bridge.md	append	post-#360 substrate sweep correction
.claude/specs/pr-d-1-fma-owl-hydrator.md	append	post-#360 substrate sweep correction
.claude/knowledge/tier-0-pattern-recognition.md	append	substrate-recognition sweep section
.claude/knowledge/pattern-recognition-cross-source.md	NEW	4-taxonomy cross-source matrix
.claude/knowledge/cca2a-sprint-prompt-template.md	NEW	substrate-grep + wrong-repo guardrail + pattern-letter discipline
.claude/board/LATEST_STATE.md	append	sprint-3 entry
.claude/board/INTEGRATION_PLANS.md	append	sprint-3 spec index

Test plan

• All 8 substrate file:symbol citations verified on main via pygithub get_contents (SchemaPtr.ontology_context_id, NamespaceRegistry::seed_defaults, BridgeFromRegistry, WoaBridge, MedcareBridge, OgitBridge, parse_ttl_directory_with_provenance, RegistryState::append ctx_id stamping)
• woa-rs#2 + medcare-rs#110 verified merged
• .grok/ 22-entry directory verified
• Both Pillars docs verified (palantir-parity-cascade-v2.md, ogit-cascade-supabase-callcenter-v1.md)
• Append-only governance preserved on all 5 files (LATEST_STATE, INTEGRATION_PLANS, PR-A-1, PR-C-1, PR-D-1, tier-0)

What you got right that's preserved

Per reviewer's "What you got right" section:

• Pattern letter discipline introduced in PR fix(knowledge): tier-0 canonical pattern letters (post-#358 correction; Pattern A is SPO-G design phase, not 'OGIT Triangle shipped') #359 — preserved + reinforced in cca2a-sprint-prompt-template
• Trivia bundle (3 quick wins) shape — preserved (W12 spec unchanged)
• .claude/board/sprint-log-N/ per-sprint coordination directory — preserved + formalized in template
• The Anti-Pattern doc itself — preserved + applied to sprint-3 itself (this PR is the meta-application)

Notes

• Same race-with-merge pattern as PR docs(architecture): unified OGIT 15-pattern synthesis + Tier-0 recognition (sprint-2) #358 → fix(knowledge): tier-0 canonical pattern letters (post-#358 correction; Pattern A is SPO-G design phase, not 'OGIT Triangle shipped') #359 → PR docs(specs): sprint-3 Tier-1 implementation specs (11 PR-X-1 + sequencing + smoke test + consumer template) #360 → fix(specs): PR-F-1 supervisor inert-bundle skip + PR-E-1 Cargo cycle (post-#360 corrections) #361. After this PR (call it docs(specs+board): post-#360 substrate-recognition sweep — 3 spec re-scopes + cross-source matrix + sprint prompt template + board hygiene #362-equivalent), the spec set is engineer-ready with substrate citations baked in.
• Future sprints follow .claude/knowledge/cca2a-sprint-prompt-template.md to prevent recurrence of all 4 documented failure modes.
• The compressed critical path (~3-4 days vs ~6 days) means sprint-4 engineer execution starts faster.

🧠 Anti-Pattern "Designing What's Already Built" applied to sprint-3 itself — recognition all the way down.

🤖 Generated by Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 220a5a43d5

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]

Correct the hydrate_once_sync reuse path

When implementing the FMA OWL/RDF-XML adapter as described here, the emitted MappingProposal and ProvenanceBundle cannot be passed to OntologyRegistry::hydrate_once_sync: I checked crates/lance-graph-ontology/src/registry.rs, and that method only accepts a ttl_root: &Path plus namespaces and then unconditionally calls parse_ttl_directory_with_provenance(...) internally. Following this spec still requires a new ingestion hook such as hydrate_from_proposals or a parser abstraction, so the "adapter calls existing hydrate_once_sync" and "only reader is new" scope will send implementers down a non-existent API path.

Useful? React with 👍 / 👎.