review

Routes code review requests to specialized review agents and aggregates their findings.

Updated May 21, 2026
One-click install
npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill review-cagesthrottleus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: review
Source: https://github.com/CagesThrottleUs/private-ai-harness/tree/main/skills/review
Command: npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill review-cagesthrottleus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Choosing and coordinating the right reviewer for each artifact is error-prone: PR diffs, specs, tests, security, CI configs, and design docs each need different checks. This Skill provides a single entry point that routes every review request to the correct specialized agent, dispatches them in parallel where possible, and merges results into one gate report. ## Core Features & Use Cases - Unified routing: One command (/review) with subcommands for PR, spec, tests, security, language, CI, HLD, API contracts, IaC, ERD, E2E, accessibility, and 15+ other review types, plus natural-language triggers like "review my PR". - Parallel multi-agent review: /review all scans the diff, conditionally dispatches pr-reviewer, spec-impl-reviewer, test-quality-reviewer, and security-reviewer in parallel, then aggregates verdicts into a merge-readiness report. - Severity-based gating: Critical findings block the merge; the aggregated report lists must-fix and should-fix items with file and line references. - Use Case: Before merging a feature branch, run /review all to get a single gate table covering PR quality, spec compliance, test quality, and security, saved to a dated report file. ## Quick Start Ask the assistant to run /review all on the current branch to dispatch all PR-scoped review agents and produce an aggregated merge-readiness report.

Frequently Asked Questions about review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a pull request with multiple specialized agents?

Run /review all to dispatch the PR reviewer, spec-implementation reviewer, test-quality reviewer, and security reviewer in parallel against the branch diff. The skill aggregates their verdicts into a single gate report with must-fix and should-fix items.

How to check if code satisfies a specification?

Use /review spec, which dispatches the spec-impl-reviewer with the spec path and diff range. It verifies the implementation actually satisfies the acceptance criteria rather than merely being annotated against them.

What review types does the /review command support?

It supports over 25 subcommands including pr, spec, tests, security, full, lang, ci, hld, api, e2e, iac, schema, accessibility, performance, and chaos. Each routes to a dedicated review agent with auto-detected input paths.

Can I trigger a code review from plain chat instead of a slash command?

Yes, direct chat phrases like "review my PR", "check my tests", or "security review" route to the corresponding subcommand automatically. The skill maps common natural-language requests to the correct review agent.

What happens when a security review finds a critical issue?

Any Critical finding from any agent escalates the overall verdict to BLOCKED, stopping the merge. If an agent blocks early, such as a missing spec, remaining agents are not run and the failure is reported immediately.

Does /review all run every available review agent?

No, /review all runs only the four PR-scoped agents: pr-reviewer, spec-impl-reviewer, test-quality-reviewer, and security-reviewer. All other agents, such as CI or HLD review, run independently through their specific subcommands.