Orchestrate a multi-agent workflow to perform a deep-dive security, correctness, and architecture review.
Scanned 9/8/2026
Install to Claude Code
npx -y skills add nobodyonlyc/skills --skill workflow-review-deep --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Workflow Review Deep?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/nobodyonlyc-workflow-review-deep)More formats (shields.io, HTML) on the badges page.
---
name: workflow-review-deep
description: Orchestrate a multi-agent workflow to perform a deep-dive security, correctness, and architecture review.
---
Deep multi-agent code review. PR number or branch: $ARGUMENTS
Multi-agent deep review: fan out one reviewer per changed file, then consolidate into a single verdict. Each phase has a detailed playbook in [`references/`](references/); see [`examples/`](examples/) for a full worked run.
## Skills this workflow drives
- [check-code-review](../check-code-review/SKILL.md) — correctness, performance, and simplification review per file (Phase 2).
- [check-security-review](../check-security-review/SKILL.md) — injection, secrets, auth, and input-validation review per file (Phase 2).
- [check-refactor](../check-refactor/SKILL.md) — shape the suggested cleanups when the user opts to apply fixes (Phase 3).
- [check-pr-review](../check-pr-review/SKILL.md) — frame the consolidated verdict as a PR review (Phase 3).
## Phases
1. **Gather & Fan-out** → [references/phase-1-gather-fanout.md](references/phase-1-gather-fanout.md)
Collect the changed files (PR or branch diff) and spawn one review agent per file (batch beyond 8).
2. **Per-file Review** → [references/phase-2-per-file-review.md](references/phase-2-per-file-review.md)
Each agent reviews its file for correctness, security, performance, and simplification.
3. **Consolidate & Verdict** → [references/phase-3-consolidate-verdict.md](references/phase-3-consolidate-verdict.md)
Merge, dedupe, sort by severity, give a verdict, and optionally apply fixes.
## Hard gates
- Sort every finding by severity (🔴 Critical / 🟡 Important / 🔵 Suggestion) before presenting.
- Give one explicit verdict: **Approve** / **Request Changes** / **Needs Discussion**, and ask before applying any fix.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!