caveman-review

Generate terse one-line code review findings from pull request diffs.

171|10|Updated Feb 20, 2026

Search Tags:#commenting #formatting #diff #code-review #pull-request #line-number

Ready to run

npx skills add https://github.com/fmflurry/settings-opencode --skill caveman-review-fmflurry

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: caveman-review
Source: https://github.com/fmflurry/settings-opencode/tree/main/skills/caveman-review
Command: npx skills add https://github.com/fmflurry/settings-opencode --skill caveman-review-fmflurry

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ultra-compressed code review comments that cut noise from PR feedback while preserving actionable signal. Each finding is a single line with location, problem, and fix. Use when a user asks for a review on a PR or diff (e.g., "review this PR", "/caveman-review"), or when it auto-triggers during pull request reviews.

Core Features & Use Cases

One-line findings: preserve exact line numbers and identifiers, with a concise issue and fix.
Flexible formatting: supports L<line> or <file>:L<line> style, with optional per-file context.
Severity prefixes: optionally include 🔴 bug:, 🟡 risk:, 🔵 nit:, ❓ q: to convey importance and urgency.
Automatic invocation: activates when a PR review is requested or the caveman-review command is used.

Quick Start

Provide terse, one-line feedback for each finding in a PR diff, preserving exact line references.

Frequently Asked Questions about caveman-review

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

FAQPage Schema

How do I generate concise code review comments from a pull request diff?▼

To generate concise code review comments from a pull request diff, this Skill analyzes the provided diff and outputs ultra-compressed, one-line findings. Each finding includes the exact line number, the identified issue, and an actionable fix.

Can I format PR review findings with specific line numbers and file prefixes?▼

You can format PR review findings with exact line references using either L<line> or <file>:L<line> style. This enforces strict formatting to preserve exact line numbers and optional file prefixes across single-file or multi-file diffs.

How do I add severity levels to automated code review feedback?▼

To add severity levels to automated code review feedback, the Skill optionally includes severity prefixes in its output. You can prefix findings with 🔴 bug:, 🟡 risk:, 🔵 nit:, or ❓ q: to convey importance and urgency.

What is the best way to reduce noise in pull request code reviews?▼

The best way to reduce noise in pull request code reviews is to use ultra-compressed feedback that cuts unnecessary text while preserving the actionable signal. This approach ensures each finding is a single line containing the location, problem, and fix.

Does this code review tool work with multi-file diffs across different programming languages?▼

This code review tool supports both single-file and multi-file diffs across various programming languages. It processes the provided pull request diff and identifies issues regardless of the language, preserving exact formatting and line references.