Pr Review logo

Pr Review

Community
dannote
pr-review

Review GitHub pull requests critically and contextually, including first reviews, re-reviews after requested changes, external reviewer or bot feedback, codebase-fit checks, validation, inline comment drafts, and merge decisions. Use when reviewing a PR, checking whether review comments were addressed, evaluating CodeRabbit or another review, preparing review comments, or deciding whether a PR is ready to merge.

Overview

Publisherdannote
Repositorydot-pi
Skill namepr-review
Stars
57
Forks
4
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by dannote on GitHub. Read the source before you install it.

Installation

Install the Pr Review AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/dannote/dot-pi/tree/master/skills/pr-review
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/dannote/dot-pi.git /tmp/dot-pi
mkdir -p .claude/skills
cp -r /tmp/dot-pi/skills/pr-review .claude/skills/pr-review
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Pr Review in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Pr Review on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Pr Review is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Pull Request Review

Review before mutating. Read the repository's AGENTS.md, CONTRIBUTING.md, PR template, and relevant bot instructions before judging the change. Treat those files as the project's source of truth; do not import conventions from another repository.

Choose the workflow

ContextRoleOutput
First reviewReviewerEvidence-based findings and one recommendation
Re-review after feedbackFollow-up reviewerStatus of each requested change plus regression findings
Merged or closed PRAuditorRetrospective findings and any separate follow-up action
CodeRabbit, Copilot, or contributor reviewMaintainerIndependently verified classifications
CI and checksReviewerAttributed pass, PR failure, infrastructure failure, flake, skip, or unknown
Preparing commentsMaintainer reviewerExact concise inline-comment draft
Approved fix or mergeMaintainer/implementerScoped action and validation result
  • First review: inspect the final diff, history, discussion, checks, and relevant source/tests.
  • Re-review: compare the reviewed commit SHA with the current head before classifying anything. Build the requested-change list from maintainer reviews, inline threads, author responses, and later commits. Mark each item as maintainer-requested, external suggestion, or newly discovered, then classify it as addressed, partial, unresolved, or obsolete from the current diff. If no commits or relevant changes followed the review, say so directly; do not infer fixes from the PR body or commit message.
  • Merged or closed PR: review retrospectively. Do not recommend merging, approving, or requesting changes; identify only lessons or a separate follow-up warranted by current code.
  • External review: verify each CodeRabbit, Copilot, or contributor finding against the current head, current PR body, current conversation, and repository policy; classify it as valid, invalid, stale, duplicate, or low-value. A bot threshold is not a project requirement unless repository guidance or CI makes it one; do not add comments or code solely to satisfy an unsupported metric.
  • CI and checks: inspect failing logs, not only check summaries. Attribute each relevant result as passing, failing because of the PR, infrastructure failure, flaky, cancelled/blocked, skipped/not applicable, or unverified. Report the actual checks; do not inflate the count with service statuses or infer checks that are not present.
  • Draft comments: map valid findings to the smallest useful line range, prefer inline comments, remove duplicates, and show the exact draft.
  • Fix or merge: only act after the user explicitly chooses that action. Use vibe-merge when reimplementing selected ideas instead of merging a PR wholesale.

Do not conflate these workflows. A request to review is not permission to comment, request changes, edit code, push, or merge.

Inspect in this order

  1. Establish the PR state, merged/closed time when applicable, base branch, worktree state, PR number, author, scope, current head SHA, and formal review decision.
  2. Read repository guidance and the PR template. Check whether guidance files differ between the base and PR branch.
  3. Read the PR title, current body, commits, timeline, formal reviews, inline comments, reactions, and issue conversation. Look for maintainer requests, external suggestions, author replies, resolved threads, and changed scope. Do not treat thanks, emoji, or reactions as approval; attribute formal approvals and change requests to their actual authors, and do not mistake a bot approval for maintainer approval.
  4. Inspect the complete diff against the correct base. Read surrounding source, analogous existing implementations, definitions of referenced helpers, and nearby tests. Resolve accessible dependencies instead of leaving findings based on assumptions.
  5. Check CI and external reviews. Inspect statusCheckRollup and failed logs rather than claiming CI is unavailable after a limited query. Separate infrastructure failures from failures caused by the change.
  6. Run focused checks and real-runtime or visual validation when the change requires it. Do not claim validation from a command you did not run.

Use gh for GitHub data. Prefer structured output and line-aware comments, for example:

bash
gh pr view <number> --comments --json title,body,author,baseRefName,headRefName,commits,reviews,comments,statusCheckRollup
gh pr diff <number>
gh api repos/OWNER/REPO/pulls/<number>/comments

Review lenses

Check only lenses relevant to the change, but always check codebase fit:

  • Correctness: behavior, edge cases, regressions, error paths, concurrency, and data loss.
  • Codebase fit: existing abstractions, package ownership, naming, file layout, dependencies, public APIs, and duplication. Search before adding a helper, type, wrapper, shim, or configuration path.
  • Tests: behavior and contracts rather than source-text matching; source-domain structure where practical; correct unit/integration/E2E/visual level; existing fixtures and helpers.
  • UX/API compatibility: documented external behavior, accessibility, framework idioms, and compatibility with the project's target platform or API.
  • Validation: relevant checks, runtime behavior, screenshots or visual diffs, and whether failures are attributable to the PR.
  • Scope and maintainability: unnecessary churn, stale compatibility code, path drilling, hand-rolled mechanisms, and misleading names or comments.

Do not reject a change merely because it differs from personal taste. Explain the repository rule, existing analogue, contract, or observed failure that makes a finding actionable.

Findings

Report findings in severity order. Each finding needs:

  • location (path:line or a precise current-diff range; retrieve the diff and verify the line before drafting);
  • concrete problem;
  • why it matters in this repository;
  • requested direction only when supported by evidence;
  • validation or reproduction, if available.

Keep the report concise. State what was checked and what was not. If there are no actionable findings, say so and list remaining validation limitations.

Comments and mutations

Default to read-only. Never post a review, inline comment, request changes, approve, edit a PR, push, or merge while merely reviewing.

Before posting comments or taking any approved action:

  1. Re-check the current head SHA. If it changed, inspect the new commits and refresh affected findings before continuing. If an existing review targets the current head and no later commits exist, state that the requested changes remain on the same reviewed code.
  2. Re-check the full conversation so the comment does not repeat an existing point or ignore an author response. Do not draft a new comment that merely repeats an unresolved maintainer request already visible on the same unchanged head unless the user explicitly wants a follow-up reminder.
  3. Prefer one concise inline comment for one actionable issue. Use a summary only for cross-cutting findings.
  4. Draft in the repository's tone: concrete, humane, and proportional. Do not expose private speculation, frustration, or unnecessary internal context.
  5. Show the exact comments and intended action. Wait for explicit approval unless the user already supplied exact final text and explicitly requested that action.
  6. After approval, post only the approved comments and verify what was posted. Do not duplicate existing comments.

Treat short follow-ups such as “go ahead” as authorization for the immediately discussed action only. If the action is unclear, summarize the pending choice instead of guessing.

Decision output

End with one recommendation appropriate to the current PR state:

  • for an open PR: approve, request changes, seek clarification or more evidence, fix selected issues ourselves, re-review after new commits, or merge only when explicitly requested and supported by checks;
  • for a merged or closed PR: no PR action, or a separate follow-up only when current code still warrants one.

Do not use missing description polish or a bot warning alone to label a contributor's work low-effort. Evaluate the code and context separately.

Frequently asked questions

What does the Pr Review AI skill do?

Review GitHub pull requests critically and contextually, including first reviews, re-reviews after requested changes, external reviewer or bot feedback, codebase-fit checks, validation, inline comment drafts, and merge decisions. Use when reviewing a PR, checking whether review comments were addressed, evaluating CodeRabbit or another review, preparing review comments, or deciding whether a PR is ready to merge.

Why use Pr Review on TypingMind?

Because you install it once and use it with any model. Pr Review is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Pr Review in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/dannote/dot-pi/tree/master/skills/pr-review. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Pr Review?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Pr Review?

As many as you like. As long as a model supports skills, you can use Pr Review with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Pr Review AI skill free?

Yes. It is published on GitHub by dannote under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇