Iterative code quality audit that loops until clean. Runs the code-steward agent in a bounded fix-audit cycle — first pass gets human approval, subsequent passes auto-fix until the audit returns clean or the iteration cap is hit.
From provenpx claudepluginhub mjmorales/claude-prove --plugin proveThis skill uses the workspace's default tool permissions.
Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.
Migrates code, prompts, and API calls from Claude Sonnet 4.0/4.5 or Opus 4.1 to Opus 4.5, updating model strings on Anthropic, AWS, GCP, Azure platforms.
Analyzes BMad project state from catalog CSV, configs, artifacts, and query to recommend next skills or answer questions. Useful for help requests, 'what next', or starting BMad.
Audit, fix, re-audit modified files, repeat until clean or capped.
test_*, *_test.*, tests/, __tests__/, *.spec.*, *.test.*, fixtures). Source first, tests adapt second..claude/.prove.json validators. Do not guess commands..prove/steward/, reports in .prove/reports/steward/.CLAUDE.md for conventions, .claude/.prove.json for validators/structure..prove/TASK_PLAN.md or .prove/plans/ for task context.$ARGUMENTS:
--full -- full codebase on pass 1 (default: changed files via git diff main...HEAD --name-only)--max-passes N -- iteration cap (default: 3).prove/steward/auto-report.md:# Auto Steward Report
**Date**: [today] **Scope**: [full | changed files | directory] **Max passes**: [N]
## Pass Log
--full or directory: use that scopegit diff main...HEAD --name-only, filter test files. No source files remaining = inform user and stop.Run the steward PCD pipeline on the determined scope. This is identical to the pipeline in skills/steward/SKILL.md Phase 1 (rounds 0a through synthesis + fallback). Produces .prove/steward/findings.md and .prove/steward/fix-plan.md.
Skip if PCD produced both files. Otherwise create .prove/steward/findings.md (standard steward format: Critical Issues, Structural Refactors, Naming & Readability, Code Hygiene, Performance, Recommendations -- numbered with file:line refs) and .prove/steward/fix-plan.md (independent, parallelizable work packages).
Present findings summary and fix plan. AskUserQuestion: "Approve all" / "Cherry-pick" / "Abort".
On abort: update auto-report, stop.
Launch parallel code-steward subagents per approved work package (findings, file list, change instructions). Serialize only packages with overlapping files.
Run .claude/.prove.json validators (lint, then tests). Note failures but do not stop the loop.
Append to auto-report:
### Pass [N]
- **Files audited**: [count]
- **Issues found**: [count] (Critical: X, Major: X, Minor: X)
- **Issues fixed**: [count]
- **Files modified**: [list]
- **Validator status**: lint [pass/fail], tests [pass/fail]
Track modified files -- they scope the next pass.
No human approval. The user approved direction in Phase 1.
Repeat until clean or capped:
Re-audit only files modified in the previous pass (excluding tests). No modified source files = loop done.
For >= 5 files: run the full PCD pipeline scoped to modified files:
python3 $PLUGIN_DIR/tools/pcd/__main__.py --project-root "$PROJECT_ROOT" map --scope <comma-separated files>
Then rounds 1-3 as in Phase 1b.
For < 5 files: skip PCD, launch code-steward directly:
Re-audit ONLY these files modified in the previous pass: [list]. Check for: (1) issues introduced by refactoring, (2) quality issues previously masked, (3) integration issues with callers/dependencies. Produce findings. Do not fix yet.
No findings = loop converges, skip to Phase 3.
Findings exist: create .prove/steward/findings-pass-N.md, auto-approve, launch fix subagents, validate and record pass (same template as 1f).
At max-passes with remaining findings: record "Cap reached" in auto-report, proceed to Phase 3.
## Summary
- **Total passes**: [N]
- **Outcome**: [Converged clean | Cap reached with N remaining issues]
- **Total issues found/fixed**: [sums]
- **Files modified**: [deduplicated]
## Test Remediation Required
| Test file | Failure reason | Source change that caused it |
|---|---|---|
## Remaining Issues (if cap reached)
[Unfixed findings from final pass]
Convergence status, per-pass breakdown, test remediation if applicable, remaining issues. If capped, suggest rerunning or manual fixes.