docs(plans): lifecycle sweep — slim shipped inventory, fix fact errors

Collapse substrate-extraction shipped tiers (1–6, partial 9–12) to stubs
with canonical pointers; slim github-marketplace (Slice 5 runbook) and
perf-triangulation (Phase 3 open work). Fix audit-worktree API name,
fts --no-fts claim, graph-CRAP motivator, apply-write-safety dupe table.

Author: SutuSebastian

docs/plans/apply-write-safety.md

@@ -44,16 +44,7 @@ Cross-file rollback on phase-2 failure; BOM round-trip policy (Q2); adversarial
 
 ## Current state (shipped)
 
-| Behavior                                    | Status                                 |
-| ------------------------------------------- | -------------------------------------- |
-| Phase-1 line match (`line content drifted`) | ✅                                     |
-| Per-file temp + `rename`                    | ✅                                     |
-| All-or-nothing on phase-1 conflicts         | ✅                                     |
-| Symlink / path-escape guards                | ✅                                     |
-| Content-hash recheck before write           | ❌                                     |
-| `fsync` before `rename`                     | ❌                                     |
-| Mixed CRLF/LF skip                          | ❌                                     |
-| Cross-file rollback on phase-2 failure      | ❌ (deferred per architecture § Apply) |
+See [`architecture.md` § Apply](../architecture.md#apply--input-modes-transport-and-policy) and the TOCTOU / EOL comments in [`apply-engine.ts`](../../src/application/apply-engine.ts). **Open (this plan):** content-hash recheck before write, `fsync` before `rename`, mixed CRLF/LF skip. Cross-file rollback stays deferred per architecture § Apply.
 
 ---
 

docs/plans/audit-delta-attribution.md

@@ -27,7 +27,7 @@ Implement **`findingKey()`** + attribution on the **`deprecated`** delta first (
 ```text
 audit --base <ref>
   → git resolve ref → sha
-  → audit-worktree.ensureCachedWorktree(sha)  →  .codemap/audit-cache/<sha>/index.db
+  → lookupCacheEntry(sha) or populateWorktree({ sha, … })  →  .codemap/audit-cache/<sha>/index.db
   → runAuditFromRef: HEAD index + base cached index
   → per delta: compute added rows → findingKey(row) vs base key Set → attribution
   → envelope.deltas[<key>].added[].attribution  (CLI / MCP / HTTP identical)

docs/plans/fts-default-on-evaluation.md

@@ -14,7 +14,7 @@
 | --- | ----------------------------------------------------------------------------------------------------------- | ------------------------- |
 | L.1 | **Measure before flipping default** — benchmark 3 fixture sizes for index time + DB bytes with/without FTS. | Empirical gate            |
 | L.2 | If adopted: change **`config.ts` default** `fts5: true` for new projects only; existing configs unchanged.  | No silent behavior change |
-| L.3 | **`--with-fts` / `--no-fts`** CLI flags continue to override.                                               | Explicit escape hatch     |
+| L.3 | **`--with-fts`** CLI flag overrides config; config `fts5: false` opts out (no `--no-fts` flag today).       | Explicit escape hatch     |
 | L.4 | Document size tax in README + packaging.md.                                                                 | Transparency              |
 
 ---

docs/plans/github-marketplace-action.md

@@ -2,7 +2,7 @@
 
 > **Status:** open · **Priority:** P2 · **Effort:** M (~2 weeks)
 >
-> **Motivator:** CRAP (Change Risk Analysis and Predictions) ranks functions that are **complex and undertested** — a standard refactor-risk metric. Codemap already has `symbols.complexity` and real `coverage` after `ingest-coverage`, but recipes like `high-complexity-untested` are empty or misleading when no coverage file exists (`COALESCE(coverage_pct, 0)` treats everything as 0%). Graph-estimated coverage uses **test-file reachability** over `dependencies` / `calls` / `test_suites` to infer coverage tiers when measured data is absent.
+> **Motivator:** CRAP ranks **complex and undertested** functions. Codemap has `symbols.complexity` + ingested `coverage`, but `high-complexity-untested` is **misleading without ingest** (`COALESCE(coverage_pct, 0)` treats missing as 0%). Graph-estimated tiers (85/40/0%) from test reachability when measured coverage is absent.
 >
 > **Roadmap:** [§ Recipe & audit enrichment](../roadmap.md#recipe--audit-enrichment)
 

docs/plans/graph-estimated-crap.md

@@ -28,7 +28,7 @@ Smallest slice: add **`mcp-tool-annotations.ts`** map + thread into **one** `reg
 MCP_TOOL_ANNOTATIONS: Record<toolName, ToolAnnotations>
   → mcp-server registerTool({ …, annotations })
   → tools/list response (advisory hints only)
-  → GET /tools parity (F.5)
+  → GET /tools parity (architecture § HTTP wiring)
 ```
 
 Apply tools already gated by `--yes` / `yes: true` in handlers — annotations are client-side UX only.