multi-agent-review — Claude Code skill: what it does, how to install
Dispatches six automated reviewers across two model tiers to audit specs and plans before execution. Flags blockers, gaps, and risks prior to code generation.
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a claude-codeWhat multi-agent-review does
This skill runs pre-execution reviews on project specifications and implementation plans. Six subagents evaluate the document across three topics: completeness, alignment, and risk. Two model tiers inspect each topic concurrently. If the fast tier and standard tier disagree on a finding, a reasoning-tier juror resolves the discrepancy.
The workflow reviews text artifacts, not source code. The coordinator validates that target files exist and that referenced mockups or companion paths resolve on disk. If the artifact exceeds roughly 2,000 lines, execution pauses to confirm whether the operator wants to inspect the entire text or focus on a specific section.
Invocations use /multi-agent-review spec or /multi-agent-review plan. Adding --fast skips the standard tier and the juror to save API tokens during rapid iteration. The final verdict categorizes items into blockers, warnings, and observations, gating subsequent implementation steps whenever blocking issues remain.
What it helps with
- catch architectural flaws in specs before implementation starts
- compare execution plans against requirement documents and mockups
- identify missing edge cases without manual document walkthroughs
- stop automated implementation when blocking issues are detected
Example prompts for the agent
- /multi-agent-review spec
- /multi-agent-review plan docs/superpowers/plans/auth.md
- /multi-agent-review spec --fast
How to install multi-agent-review
1Skills CLI (Vercel)
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a claude-code2Manually: put the files in place
~/.claude/skills/multi-agent-review/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/multi-agent-review/
1Skills CLI (Vercel)
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a codex2Manually: put the files in place
.agents/skills/multi-agent-review/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a cursor2Manually: put the files in place
.cursor/skills/multi-agent-review/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install riekelt/multi-agent-review multi-agent-review2Manually: put the files in place
.github/skills/multi-agent-review/SKILL.md1Skills CLI (Vercel)
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a gemini-cli2Platform CLI
gemini skills install https://github.com/riekelt/multi-agent-review --path plugins/multi-agent-review/skills/multi-agent-reviewManually: put the files in place
.devin/skills/multi-agent-review/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add riekelt/multi-agent-review --skill multi-agent-review -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about multi-agent-review
How to install multi-agent-review in Claude Code?
Run npx skills add riekelt/multi-agent-review --skill multi-agent-review -a claude-code in your terminal.
What does multi-agent-review do?
It conducts a multi-agent review of technical specifications and execution plans across completeness, alignment, and risk before any code is written.
Does multi-agent-review inspect code files?
No, it strictly reviews markdown specifications and planning documents. It does not perform source code review.
