# Pr Review Fix Loop

> |

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

## Install

```sh
agentstack add skill-dapi-claude-code-marketplace-pr-review-fix-loop
```

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

## About

# PR Review Fix Loop

Navigator for iterative PR review + autofix loop. Analyzes context and suggests the optimal command.

## When to activate

- User wants to review their PR changes
- User wants to fix review comments automatically
- User mentions "review loop", "autofix", "check PR"
- User wants Codex CLI review (standalone)

## Instructions

### Step 1: Determine user intent

Determine which command fits:

| Intent | Command |
|-|-|
| Iterative review + fix cycle | `/pr-review-fix-loop` |
| Standalone Codex review | `/codex-pr-review` |
| Just review without fixing | Suggest `/pr-review-toolkit:review-pr` instead |

If intent is unclear, ask the user.

### Step 2: Suggest parameters (for /pr-review-fix-loop)

Based on context, suggest relevant flags:

- If user mentions specific aspects (code, errors, tests, types, comments, simplify) -- add `--aspects "..."`
- If user mentions criticality threshold -- add `--min-criticality N`
- If user mentions linting -- add `--lint`
- If user mentions Codex -- add `--codex`
- If user mentions specific base branch -- add `--base BRANCH`
- If user mentions iteration limit -- add `--max-iterations N`

Default (no flags) is fine for most cases.

### Step 3: Present the command

Show the user the suggested command with brief explanation:

```
/pr-review-fix-loop [flags]
```

Explain what will happen:
1. PR review via pr-review-toolkit
2. Automatic fix of issues above criticality threshold
3. Repeat until clean or stagnation detected
4. Report saved to `.claude/pr-review-loop-report.local.md`

### Step 4: Let user run the command

Do NOT run the command yourself. Present it and let the user invoke it.

## Examples

### Example 1: Simple PR review

User says: "review my PR"
Action: Suggest `/pr-review-fix-loop`
Result: User runs the command, loop starts

### Example 2: Review with specific focus

User says: "check my PR for errors and code issues only"
Action: Suggest `/pr-review-fix-loop --aspects "code errors"`

### Example 3: Review with Codex

User says: "review my changes with codex too"
Action: Suggest `/pr-review-fix-loop --codex`

### Example 4: Codex-only review

User says: "run codex review"
Action: Suggest `/codex-pr-review`

### Example 5: Fix review comments

User says: "fix the review comments on my PR"
Action: Suggest `/pr-review-fix-loop --min-criticality 3` (lower threshold to catch more issues)

## Troubleshooting

### Plugin not installed
If `/pr-review-fix-loop` command is not available, tell user:
```
/plugin install pr-review-fix-loop@dapi
```

### Required plugins missing
If pr-review-toolkit or feature-dev not installed:
```
/plugin install pr-review-toolkit
/plugin install feature-dev
```

## Source & license

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

- **Author:** [dapi](https://github.com/dapi)
- **Source:** [dapi/claude-code-marketplace](https://github.com/dapi/claude-code-marketplace)
- **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-dapi-claude-code-marketplace-pr-review-fix-loop
- Seller: https://agentstack.voostack.com/s/dapi
- 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%.