# Ralph Review

> Critically review a current or completed Ralph sprint for specification alignment, implementation correctness, validation evidence, orchestration reliability, and remaining gaps. Use when the user asks to review, audit, accept, or assess a Ralph sprint. Do not implement fixes unless the user separately authorizes them.

- **Type:** Skill
- **Install:** `agentstack add skill-zacharygcook-agent-skills-ralph-review`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [zacharygcook](https://agentstack.voostack.com/s/zacharygcook)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [zacharygcook](https://github.com/zacharygcook)
- **Source:** https://github.com/zacharygcook/agent-skills/tree/master/skills/ralph-review

## Install

```sh
agentstack add skill-zacharygcook-agent-skills-ralph-review
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Ralph Review

Review the sprint as an evidence-backed delivery unit.

1. Read repository instructions, the durable spec, sprint `README.md`, `IMPLEMENTATION_PLAN.md`,
   `relevant-specs.md`, `chunks.json`, and `SCRATCHPAD.md`.
2. Inspect the actual commit range and changed artifacts. Do not accept summaries as proof.
3. Verify every passed chunk against its acceptance criteria, validation logs, and commit evidence.
4. Check final review, documentation, sprint-validation, and optional E2E hook states. Distinguish
   skipped, failed, interrupted, and completed hooks.
5. Look for missing spec behavior, accidental scope, weak tests, stale documentation, unsafe
   orchestration state, and negative knowledge the next sprint must preserve.
6. Report findings by severity, then give a clear verdict: complete, repairable before acceptance, or
   blocked. Name the exact evidence and next action.

Do not rewrite chunk state, manufacture evidence, or implement findings during a review-only request.
When installed beside `$ralph-loop`, consult `references/review.md` for the full shared rubric.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [zacharygcook](https://github.com/zacharygcook)
- **Source:** [zacharygcook/agent-skills](https://github.com/zacharygcook/agent-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-zacharygcook-agent-skills-ralph-review
- Seller: https://agentstack.voostack.com/s/zacharygcook
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.