feat: repo-agnostic refactor (BMad spec-test-build loop)

- NEW: repo-profiler.js — deterministic archetype detection (Infra, Frontend, Backend, etc.)
- NEW: extract-dynamic.js — generic extractor replacing hardcoded Foxtrot patterns
- NEW: eval-generator.js — dynamic ground-truth question generation from any repo graph
- NEW: specs/bmad-agnostic-refactor-spec.md — full BMad spec with acceptance criteria
- REFACTORED: prose.js — two-pass LLM synthesis with rich context (shared secrets, ports, service refs)
- REFACTORED: sysdoc.js — wired repo-profiler + extract-dynamic, --legacy escape hatch
- REFACTORED: wiggum-v2.sh — uses eval-generator before benchmarks
- FIXED: graph.js — _edgeSet rebuilt on loadSnapshot() (edge dedup was broken)
- FIXED: graph.js — recursive sortKeys() for deep equality in diffing
- FIXED: prose.js — robust JSON array extraction from LLM output
- FIXED: ratchet.js — syntax validation (node --check) before saving LLM mutations
- FIXED: extract-dynamic.js — centralized state services regex, added console.warn for silent failures
- TESTS: test-eval-generator, test-repo-profiler, test-synthesis-quality + mock fixtures

Eval: 81.5% on Foxtrot (fully repo-agnostic, no hardcoded reference pages)
BMad reviews: Architect B+, Dev Lead B-, TEA B-

wiggum-v2-bmad.log

@@ -0,0 +1,49 @@
+🔁 Ralph Wiggum Loop (V2) — max 1 iterations, target 77%
+Benchmark: Confluence Gold Standard (/home/node/.openclaw/workspace/projects/dev-intel-v2/eval-confluence-ref-questions.json)
+
+=== Iteration 1/1 ===
+📝 Running V2 pipeline...
+  State services: 9
+Auto-detected 0 entry points
+Flow traces: 0 valid out of 0 attempted
+Impact analysis: 0 nodes with downstream dependents
+Generating architecture overview...
+Generating prose for subsystem: account-common...
+Generating prose for subsystem: app-common...
+Generating prose for subsystem: app-tools...
+Generating prose for subsystem: compute-common...
+Generating prose for subsystem: compute-tools...
+Generating prose for subsystem: control-core...
+Generating prose for subsystem: ipam-core...
+Generating prose for subsystem: ipam-tools...
+Generating prose for subsystem: network-common...
+Generating prose for subsystem: network-core...
+Generating prose for subsystem: runtime...
+Generating prose for subsystem: root...
+Generating prose for 124 contracts...
+Agent KB: 12 subsystems, 76 charts
+Synthesizing dynamic reference pages via LLM...
+Dynamic reference pages and index synthesized.
+Generated docs in ./foxtrot-docs
+- 12 subsystems
+- 124 contracts
+- 0 flows
+📊 Running agent file-browsing eval against Confluence questions...
+Using model: claude-haiku-4.5
+Agent Eval: 32 machine-audience questions
+[1/32] arch-layered-order... 25% (A:1 C:1 P:2 N:1) files:2
+[2/32] arch-hub-spoke-ownership... 70% (A:4 C:2 P:4 N:4) files:2 [NOT_FOUND]
+[3/32] arch-aws-regions... 50% (A:2 C:5 P:1 N:2) files:2
+[4/32] arch-gcp-shared-vpc-host... 45% (A:2 C:1 P:4 N:2) files:2 [NOT_FOUND]
+[5/32] arch-cidr-employee-access... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[6/32] arch-production-cidr... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[7/32] dep-runtime-common-horizontal... 20% (A:0 C:0 P:3 N:1) files:2 [NOT_FOUND]
+[8/32] dep-vertical-layers... 45% (A:2 C:1 P:3 N:3) files:2
+[9/32] dep-create-account-repos... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[10/32] dep-create-cluster-repos... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[11/32] dep-compute-common-deps... 15% (A:1 C:0 P:1 N:1) files:2
+[12/32] ops-argocd-deployment-flow... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[13/32] ops-ebf-release-pattern... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[14/32] ops-rollback-procedure... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[15/32] ops-branch-cluster-mapping... 25% (A:0 C:0 P:5 N:0) files:2 [NOT_FOUND]
+[16/32] ops-jenkins-jobs...