MCPInstall

coderabbit

This skill should be used when the user asks to "triage coderabbit", "check coderabbit comments", "analyze coderabbit review", "fetch coderabbit findings", "coderabbit PR", "review coderabbit suggestions", "run coderabbit review", "coderabbit review", "review with coderabbit", or mentions analyzing CodeRabbit PR comments or running CodeRabbit CLI review locally.

Installs
120
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/paulrberg/agent-skills --skill coderabbit
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads coderabbit automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

CodeRabbit Review Triage

Fetch and analyze CodeRabbit review comments from a GitHub PR, or run a local CodeRabbit CLI review. Default mode fetches from the remote PR. Use --local to run the CLI against local diffs.

Arguments

Parse $ARGUMENTS for:

  • <pr-number> — integer PR number
  • <pr-url> — GitHub PR URL (e.g., https://github.com/owner/repo/pull/123), extract the PR number from the path
  • --local — switch to local review mode (see below)

Mode Dispatch

If $ARGUMENTS contains --local, strip it and forward remaining arguments to references/local-review.md. Follow that workflow instead of continuing here. All flags after --local (--base, --base-commit, --type, --config) are forwarded to the local workflow.

Otherwise, continue with the PR comment analysis workflow below.

Prerequisites

Run these checks in order. Stop at the first failure.

1) GitHub CLI Authenticated

gh auth status

If not authenticated, stop and tell the user to run gh auth login.

2) Detect Repository

gh repo view --json owner,name --jq '"\(.owner.login)/\(.name)"'

Store the result as {owner}/{repo}.

3) Resolve PR Number

  • If a PR number was provided as an argument, use it directly.
  • If a PR URL was provided, extract the number from the URL path.
  • If neither was provided, auto-detect from the current branch:
gh pr view --json number --jq '.number' 2>/dev/null

If auto-detection fails, stop: "No open PR found for the current branch. Provide a PR number or URL, or use --local for local review."

4) Validate PR

gh pr view {pr_number} --json state,url --jq '"\(.state) \(.url)"'

If the command fails, the PR does not exist. Report and stop. Otherwise, note the state (OPEN/MERGED/CLOSED) and URL for the report scope section.

Workflow

1) Fetch CodeRabbit Comments

Fetch from three GitHub API endpoints, filtering for the CodeRabbit bot. Use --paginate for large PRs. The bot's login is coderabbitai but may appear as coderabbitai[bot], so match with startswith.

Walkthrough/summary comments (issue comments where CodeRabbit posts its walkthrough):

gh api "repos/{owner}/{repo}/issues/{pr_number}/comments" \
  --paginate \
  --jq '[.[] | select(.user.login | startswith("coderabbitai"))]'

Review objects (top-level review body and state):

gh api "repos/{owner}/{repo}/pulls/{pr_number}/reviews" \
  --paginate \
  --jq '[.[] | select(.user.login | startswith("coderabbitai"))]'

Inline review comments (file-level comments with path and line info):

gh api "repos/{owner}/{repo}/pulls/{pr_number}/comments" \
  --paginate \
  --jq '[.[] | select(.user.login | startswith("coderabbitai"))]'

If no CodeRabbit comments are found across all three endpoints, report "No CodeRabbit review found on PR #{pr_number}" and stop.

2) Parse Comments into Findings

Process the raw API responses into a normalized list of findings:

  • From inline comments (pulls/{pr}/comments): Each comment becomes a finding with path (from .path), line (from .line or .original_line), body (the comment text), diff_hunk (surrounding context), and source: "inline". These are the primary source of actionable findings.
  • From review bodies (pulls/{pr}/reviews): Parse each review's body for actionable items. Extract individual suggestions. Mark source: "review".
  • From issue comments (issues/{pr}/comments): Parse for file-level summaries and actionable items within the walkthrough. Mark source: "walkthrough".

Inline comments are the primary source of actionable findings. Review summaries and walkthrough comments provide context and may surface additional actionable items not tied to specific lines.

3) Triage Findings

Load references/triage.md and follow the shared triage process — categorize, classify, confirm ambiguous, generate fix plan, and report.

Stop Conditions

Stop and ask for direction when:

  • gh CLI is not authenticated.
  • The PR does not exist or is not accessible.
  • No CodeRabbit comments found on the PR.
  • Auto-detection fails and no PR number was provided.
  • A finding requires architectural changes beyond the PR's scope.
  • Classification confidence is low for code on critical paths (auth, payments, data integrity).

FAQ

What does the coderabbit skill do?

This skill should be used when the user asks to "triage coderabbit", "check coderabbit comments", "analyze coderabbit review", "fetch coderabbit findings", "coderabbit PR", "review coderabbit suggestions", "run coderabbit review", "coderabbit review", "review with coderabbit", or mentions analyzing CodeRabbit PR comments or running CodeRabbit CLI review locally.

How do I install coderabbit?

Run npx skills add https://github.com/paulrberg/agent-skills --skill coderabbit in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/coderabbit/ (personal) or .claude/skills/coderabbit/ (project).

Which AI agents can use coderabbit?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is coderabbit safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains coderabbit?

It is published by paulrberg in paulrberg/agent-skills. It has 120 recorded installs.

More from paulrberg/agent-skills

View all →

effect-ts

paulrberg/agent-skills

Use for nontrivial Effect 4 work including services/layers, typed errors, Schema/JsonSchema, Config, runtime/concurrency, @effect/vitest, effect/ai and @effect/ai-, effect/sql and @effect/sql-, Effect Atom (@effect/atom-), or @prb/effect-next.

Data & Databases3.1K installs

tailwind-css

paulrberg/agent-skills

Use for Tailwind v4 styling: add/fix classes, configure or migrate Tailwind, use tailwind-variants, or tw-animate-css.

cli-gh

paulrberg/agent-skills

Use for GitHub CLI automation: repository reads, workflow runs, search, codespaces, releases, configuration, or gh command syntax. Use yeet for contribution writes to PRs, issues, comments, or discussions.

code-polish

paulrberg/agent-skills

Polish changed code when the user explicitly asks, or when an active workflow requests post-implementation simplification and risk-profiled review over a fixed file scope.

AI & Agents2.8K installs

yeet

paulrberg/agent-skills

Use for GitHub PR/issue/discussion workflows: create/update PRs, issues, or discussions and post issue or discussion comments; triggers include yeet.

AI & Agents2.6K installs

bump-release

paulrberg/agent-skills

Cut a release: bump versions, write changelogs, commit, tag.

AI & Agents2.4K installs

Popular code quality & workflow skills

View category →

grill-me

mattpocock/skills

A relentless interview to sharpen a plan or design.

Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

handoff

mattpocock/skills

Compact the current conversation into a handoff document for another agent to pick up.

domain-modeling

mattpocock/skills

Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing a GLOSSARY.md, or recording or editing an ADR.

codebase-design

mattpocock/skills

Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.

diagnosing-bugs

mattpocock/skills

Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.