devpilot-scanning-reposlisted

# Repo Scan (Security / Edge Cases / Coverage → GitHub Issues) ## Files in this skill | File | When to load | |---|---| | `agents/security-scanner.md` | Step 3 — sub-agent prompt for the security scanner. | | `agents/edge-case-hunter.md` | Step 3 — sub-agent prompt for edge-case hunting (no business logic). | | `agents/coverage-auditor.md` | Step 3 — sub-agent prompt for test-coverage gap detection. | | `agents/doc-consistency-auditor.md` | Step 3 — sub-agent prompt for doc/code drift detection (CLAUDE.md, AGENTS.md, README.md and linked docs). | | `references/scoring.md` | Step 4 — full 0/25/50/75/100 rubric + false-positive classes. | | `references/issue-template.md` | Step 7 — exact `gh issue create` body and label contract. | | `references/labels.md` | Step 2 — one-shot `gh label create` commands. | | `scripts/check-findings.py` | Step 3.5 — validates each scanner's JSON output against the schema. | | `evals/evals.json` | Test scenarios for skill behavior (not loaded at runtime). | ## Overview A whole-repo sweep that dispatches **four parallel specialist sub-agents** (security, edge-case, coverage, doc-drift), scores every finding 0–100 for confidence, filters below threshold, then files each surviving finding as a labeled GitHub issue. Business logic is out of scope — scanners only catch mistakes a reasonable reader could flag without domain knowledge. The doc-drift scanner audits the entry-point docs (`CLAUDE.md`, `AGENTS.md`, `README.md`) and every doc file they li