fix(skills): green the skill-flow gate after skill-prune Phase 2

[boshu2]

Why

main is RED on the validate-skill-flow gate (wired into the contracts-sync CI job). The skill-prune Phase 2 (ag-if7p) deleted six skills — operating-loop-skill, brainstorm, design, ratchet, complexity, cass-memory — but left dangling references to them in surviving skills, producing 16 hard findings:

• 6 × consumes-vocabulary — agy-native, discovery (×2), eval-outcomes, refactor, rpi consume a deleted skill
• 8 × metadata.dependencies — acfs, discovery (×2), eval-outcomes, perf, refactor, rpi, test depend on a deleted skill
• 2 × orphan — beads-bv, cc-hooks are new skills with zero edges and no allowlist entry

What

• Purge every reference to the six pruned skills across all three edge layers (consumes, metadata.dependencies, context_rel) in the 9 affected skills. Verified each survivor keeps its remaining edges — no new orphans introduced.
• Allowlist the two genuine leaf skills: beads-bv (tracker-triage boundary to bv/br) and cc-hooks (Claude Code hooks runtime-config boundary).
• Regenerate docs/contracts/context-map.md.

Verification

$ bash scripts/validate-skill-flow.sh
validate-skill-flow: 67 skill(s), 96 skill-to-skill edge(s)
  orphans: 10 (allowlisted standalone: 10)
OK: skill flow is connected and the consumes vocabulary is closed.   # exit 0, zero hard findings

• validate-skill-flow.sh → exit 0 ✅
• validate-context-map-drift.sh → exit 0 ✅
• All 9 edited frontmatters parse as valid YAML ✅

Out of scope (separate follow-up)

The gate still emits two non-fatal WARNs left by the four-vendor-corpus (#803) and prune PRs: ~45 stale allowlist entries (dead refs to pruned skills) and ~60 unused EXTERNAL_INPUTS tokens. These don't block CI; cleaning them (and optionally escalating those WARNs to FAIL so they can't re-rot) is a deliberate follow-up, not bundled here to keep this fix-forward tightly scoped to greening the gate.

Bounded-context: BC1-Corpus
Evidence: OK: skill flow is connected and the consumes vocabulary is closed.

---

Generated by Claude Code

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_942bff22-3a15-4e0f-b09b-c321ac2aae15)