Menu
Batch-apply findings from the most recent --local review. One confirmation, one test run.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Batch-verify and delete dead code: rg-confirm zero callers for N symbols, delete files, run the build once, stage results.
Run benchmarks and check regressions, add benchmark gating to CI, or audit GitHub Actions workflows. Determines mode from user intent.
Orchestrated code review of git diff using parallel agents. Standard mode (diff-scoped) or full mode (adds Roslyn semantic analysis). Produces severity-grouped plan.
Unified code review: --local (diff review), --pr (adversarial PR review), --apply (batch fixes from last review).
Audit .NET solution health across 13 dimensions: dependency graph, solution sync, CPM compliance, package health, framework consistency, IVT & encapsulation, build config, analyzer config, NuGet config, suppression hygiene, output type & AOT, public API surface, and lock files & pruning. Works with any .NET solution (.slnx or .sln). Parses all config files once, spawns 3 parallel agents, and produces a severity-grouped report with a Mermaid dependency graph. Triggers on: solution audit, audit solution, project graph, dependency audit, solution health, audit projects, check solution.
Scans source and .claude/ files for anti-pattern violations, applies the top mechanical fixes, and reports pre/post counts.
| name | review-apply |
| description | Batch-apply findings from the most recent --local review. One confirmation, one test run. |
| when_to_use | dispatched by review router |
| argument-hint | [--include-low] |
| user-invocable | true |
Batch-apply findings from the most recent --local review. One confirmation, one test run.
Tasks.md (preferred): Find the latest review plan:
fd -t d 'review-' plans/ | sort -r
Read its tasks.md — parse each - [ ] \{file}:{line}`entry. Group by file. Skip LOW unless--include-low`.
If no review plan found → fall back to context.md:
fd -e md . plans/ -t f --regex 'review-\d{4}-\d{2}-\d{2}.*/.*context\.md' | sort -r
If neither exists → report: "No saved review found. Run review --local first."
Parse findings grouped by severity. Skip: needs-verification, advisory-only, timed-out agents, known exceptions (see references/known-exceptions.md).
Map each finding → exact file + line + planned edit. Group by file. If >10 files, warn "Large blast radius ({N} files)." Verify accuracy against current code (code may have changed since review).
Display confirmation table. Auto-apply CRITICAL + HIGH + MEDIUM. Skip LOW unless --include-low.
Read all affected files first (parallel Reads), then apply all edits. Group findings by file, apply all in one Edit per file. Hold the test run until every fix is applied.
Cascade stop rule: If >5 test failures in files outside the edit plan, stop and report the cascade. Don't chase it.
AskUserQuestion before: public method/interface changes, >10 line deletions, test expectation changes.
Spawn build-validator agent for changed files.
Skip if Phase 4 passed cleanly and all fixes were mechanical. Cap N=2. Self-review → verify → cross-review per iteration.
Applied Findings
================
N findings applied across M files
Targeted tests: {N passed, M failed}
Files modified: ...
Not applied (manual review needed): ...