# Harness Review > Use when the user asks to review work done with the Harness framework, mentions /review, or wants changes checked against AGENTS.md, docs/ARCHITECTURE.md, docs/ADR.md, docs/PRD.md, and docs/UI_GUIDE.md. This skill prioritizes findings, checks architecture and quality, and can use codex review when useful. - Skill: `tomevault-io/harness-review` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/harness-review` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/harness-review/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Product & Planning - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/harness-review --- # Harness Review Review code against the project docs and Harness rules. ## Read First 1. `AGENTS.md` 2. `docs/ARCHITECTURE.md` 3. `docs/ADR.md` 4. `docs/PRD.md` 5. `docs/UI_GUIDE.md` if UI is involved ## Review Focus Prioritize: - Regressions - Spec mismatches - Missing tests - Security and data integrity risk - Architecture drift ## Process 1. Identify the change set. 2. Review the diff or changed files. 3. Compare the result against project docs. 4. Report findings first, ordered by severity. When helpful, use: ```bash codex review --uncommitted ``` or compare against a base branch with: ```bash codex review --base main ``` ## Checklist - Follows `AGENTS.md` guardrails - Matches `docs/ARCHITECTURE.md` - Respects ADR decisions - Supports PRD requirements - Includes tests for new behavior - Avoids banned UI patterns from `docs/UI_GUIDE.md` ## Output - Findings first - Then open questions or assumptions - Then a short summary only if needed If no findings remain, say that clearly and mention any residual test gap. --- > Source: [soohy00/codex-harness-framework](https://github.com/soohy00/codex-harness-framework) — distributed by [TomeVault](https://tomevault.io).