100%

Walking Skeleton — dead-code end-to-end

Goal

Prove the FULL graph-driven pipeline on ONE rule (WASTE:dead-code) end-to-end before widening: rules-JSON detection descriptor -> engine reads it -> runs the dead-code cypher -> verify-gate -> finding -> golden probe on the real /impl-hygiene-review entry point. dead-code is already graph-native (is_dead_code_candidate GDS property + existing seam_scan query), so the skeleton de-risks the NEW pipeline, not unbuilt data.

Implementation Sketch

Thinnest vertical slice: add a minimal detection: descriptor for WASTE:dead-code, a minimal descriptor-driven engine path that runs the existing seam_scan dead-code detector through the verify-gate to the report sink, and a golden probe asserting the graph-emitted dead-code findings match a hand-labeled ori_arc subset. Wire the real skill entry point (production-path, not seam-injected).

Spec References

seam_scan.py compose_dead_code_cypher; verification_gate.py; finding_emit.py; .claude/rules/intelligence.md Graph-First Mandate; impl-hygiene-rules.json WASTE:dead-code.

Work Items

  • Add a minimal detection: descriptor (tier=graph-native, detector=dead-code, graph_props=[is_dead_code_candidate]) to WASTE:dead-code in impl-hygiene-rules.json; schema-validate.
  • Thin descriptor-driven engine path: read the descriptor, run seam_scan dead-code detector, route every candidate through verify_gate (drop-unless-verified), emit via the report sink.
  • Hand-label a dead-code golden subset for ori_arc; golden probe asserts graph-emitted findings match it (recall) with zero per-file LLM walk.
  • Wire the real /impl-hygiene-review entry point to the skeleton path for dead-code (production-path end-to-end, not a seam-injected stub).

Fresh intel (regenerated)

{ “schema_version”: 1, “target”: { “kind”: “plan-section”, “ref”: “graph-driven-hygiene/s-5525f25d” }, “generated_at”: “2026-06-21T21:36:34.876307+00:00”, “graph_state”: { “head_sha”: “194a9803”, “last_code_import_at”: “2026-06-21T21:31:40.514Z”, “embedding_stale”: false, “insights_stale”: false, “cpg_stale”: false }, “surfaces”: {}, “summary”: “intel-package[plan-section:graph-driven-hygiene/s-5525f25d] agent-authored dossier”, “degraded”: false, “dossier”: { “objective_symbols”: [], “tiers”: {}, “agent_authored”: true, “difficulty”: { “class”: “routine”, “research_online”: false, “research_mode”: “auto”, “signals”: [], “plan_dir”: “/home/eric/projects/ori_lang/plans/graph-driven-hygiene” } } }

(full dossier: walking-skeleton—s-5525f25d.intel.json)