spec-quality-reviewlisted

## What It Does Runs a structural review over a spec to find contradictions, formula inconsistencies, boundary gaps, goal conflicts, dangling references, and missing negative cases before the spec is treated as authoritative. ## Use This When Use this before `spec-compliance-extract`, after a spec edit, or when an implementation bug suggests the spec may be internally inconsistent. ## Inputs - Read the target spec markdown file in full. - Read `references/spec-defect-checklist.md` before issuing findings. - Reuse the persisted `.paqad/compliance/<spec-slug>/spec-review.json` report when it exists so resolved findings can be carried forward. - Use `agents/openai.yaml` for the reasoning contract when the workflow executes through agent infrastructure. ## Procedure 1. Run `scripts/scan-defects.sh <spec.md>` — it skips Open Questions and TBD lines and emits candidate hits across the categories in `assets/categories.txt` (vague-quantifier, missing-actor, unbounded-modal, tbd-leak, dangling-ref, goal-collision, missing-negative). 2. Confirm each hit (or dismiss with reason) — the script flags candidates, the LLM judges severity. 3. Persist the machine-readable report under `.paqad/compliance/<spec-slug>/spec-review.json`. 4. Format markdown summary per `assets/output.template.md`; validate with `scripts/lint-output.sh`. 5. Order findings by severity, then by source line. ## Output Contract - Return a `Findings` section summarizing active defects with severity, category, lo