caveman-review

Generates terse one-line code review comments with location, problem, and fix.

Updated Jan 28, 2026
One-click install
npx skills add https://github.com/ttphats/project-detedxs26 --skill caveman-review-ttphats
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: caveman-review
Source: https://github.com/ttphats/project-detedxs26/tree/main/.augment/skills/caveman-review
Command: npx skills add https://github.com/ttphats/project-detedxs26 --skill caveman-review-ttphats

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code review feedback is often bloated with hedging, filler phrases, and vague suggestions that slow down authors and reviewers. This Skill compresses every review finding into a single actionable line so pull request feedback stays fast and unambiguous. ## Core Features & Use Cases - One-Line Findings: Each comment follows the format L<line>: <problem>. <fix>. with exact line numbers and symbol names in backticks. - Severity Prefixes: Optional tags like bug:, risk:, nit:, and q: classify findings so authors can triage quickly. - Anti-Hedging Rules: Strips filler phrases, restatements, and hedging language while keeping concrete fixes and the reasoning behind non-obvious changes. - Use Case: When reviewing a multi-file pull request diff, invoke the skill to produce compact per-line comments such as auth.ts:L42: bug: user can be null after .find(). Add guard before .email. ## Quick Start Ask the assistant to review this PR or diff using caveman-review to get terse one-line findings.

Frequently Asked Questions about caveman-review

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

FAQPage Schema
How do I write concise code review comments?▼

Use a one-line format per finding: line number, the problem, and the concrete fix. Drop filler phrases like "I noticed that" and hedging words like "perhaps", and reference exact symbol names in backticks so authors can act immediately.

How to review a pull request diff quickly?▼

Invoke the skill with a phrase like "review this PR" or "/caveman-review" and it produces terse per-line findings across all changed files. Multi-file diffs use the format file:line followed by problem and fix.

What severity labels should code review comments use?▼

The skill supports four optional prefixes: bug for broken behavior, risk for fragile code like races or missing null checks, nit for style issues authors can ignore, and q for genuine questions rather than suggestions.

When should code review feedback not be terse?▼

Terse mode is dropped for security findings such as CVE-class bugs that need full explanation, architectural disagreements requiring rationale, and onboarding contexts where reviewers need more teaching-oriented detail.

Does this skill write the code fixes it suggests?▼

No. The skill only produces review comments and does not write code fixes or approve or request changes on a pull request. Saying "stop caveman-review" or "normal mode" reverts to standard behavior.