New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Review PR

by pymc-labs in pymc-labs/CausalPy

Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns.

Apache-2.0Auto-check passedDevelopment

Install Review PR

skills CLI
$ npx skills add pymc-labs/CausalPy --skill review-pr -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install pymc-labs/CausalPy review-pr --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/pymc-labs/CausalPy.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/review-pr .claude/skills/review-pr && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
review-pr
GitHub stars
1.2k
Token cost
~2.2k tokens
SKILL.md length
1,021 words
Files
13
Skills in repo
12
Repo updated
First seen
Licence
Apache-2.0

At a glance

Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns.

  • Works in 7 steps: Identify the PR number or current… → Inspect the local working tree before… → Fetch PR metadata, commits, reviews,… → …
  • Asked to review a PR
  • SKILL.md covers Boundary, Review Stance, Intake and Classify the PR, plus 4 more sections
  • Calls gh and uv

What it does

Review PR is an agent skill from pymc-labs/CausalPy. Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns. Use when asked to review a PR, assess a branch before merge, summarize PR risks, or request changes.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 13 other files (for example `resources/code-patterns.md`, `resources/docs-patterns.md` and `resources/maintenance.md`).

It sits in Development, covering Pull requests and Type safety. The repository describes itself as: A Python package for causal inference in quasi-experimental settings. The licence is Apache-2.0.

When your agent uses it

  • Asked to review a PR
  • Assess a branch before merge
  • Summarize PR risks
  • Request changes

Example prompts

  • “/review-pr”

Workflow steps

7 steps, taken from the first numbered list in SKILL.md.

  1. Identify the PR number or current branch, base branch, head branch, and whether the branch tracks a remote.
  2. Inspect the local working tree before any git operation and preserve unrelated local changes.
  3. Fetch PR metadata, commits, reviews, issue comments, changed files, mergeability, and check summary.
  4. Check whether the branch is behind its base and whether GitHub reports conflicts. Do not resolve conflicts as part of review unless…
  5. Check remote CI with gh pr checks or the equivalent GitHub command. Distinguish failed, pending, skipped, and missing required checks.
  6. Inspect the full PR diff against the base branch, not only the latest commit.
  7. Verify contributor claims against code, tests, and branch history before accepting them.

What it can do on your machine

Read from SKILL.md and the folder at commit 7882153. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • gh
    • uv

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use gh and uv, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Review PR loads about 2.2k tokens when it runs. Until then it costs about 72 tokens; SKILL.md has 1,021 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~72
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from pymc-labs/CausalPy at commit 7882153, republished under its Apache-2.0 licence (© pymc-labs). 1,021 words, ~2,186 tokens.

Download SKILL.mdSave it as .claude/skills/review-pr/SKILL.md (or your agent's skills folder). This skill also uses 12 other files; get the full folder from GitHub.
name
review-pr
description
Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns. Use when asked to review a PR, assess a branch before merge, summarize PR risks, or request changes.

Maintainer PR Review

Use this skill to evaluate whether a PR is correct, safe, understandable, and merge-ready. This is a review workflow, not primarily a fix workflow.

Boundary

  • If the user asks to review, assess, summarize risks, or decide whether a PR is ready to merge, use this skill.
  • If the user asks to make the PR green by fixing CI, conflicts, or review comments, use pr-to-green for CausalPy-specific greening work.
  • For continuous merge-readiness monitoring, repeat this skill's intake, CI, and comment checks on a cadence. If the repository later adds a dedicated monitoring skill, prefer that.
  • If the review uncovers clear, small fixes and the user asked you to fix them, keep changes scoped to the PR's intent and follow the repo's commit and prek rules.
  • Never post review comments, approve, request changes, or merge through GitHub without explicit human approval.
  • Do not duplicate mechanical checks already covered by hooks and CI. If a recurring issue is mechanically enforceable but not enforced, recommend a follow-up issue instead of treating each instance as bespoke review work.

Review Stance

A review must actively try to falsify the PR's claims; passing CI is evidence, not the conclusion. Before reading the whole diff, state a private review brief with:

  • The user outcome and two to four concrete wins the PR intends to deliver.
  • The changed contracts, assumptions, and public surfaces most likely to regress.
  • At least three PR-specific failure hypotheses, each paired with the code path, caller, test, or runtime behavior that could disprove it.

For every changed production behavior, trace both directions across its seam: its inputs/producers and its outputs/consumers. Compare old and new behavior at the most fragile boundary (multi-output, empty or degenerate data, optional backend, custom subclass, serialization, or backwards-compatible call) rather than only the happy path. Read existing PR comments as review leads, not as conclusions. Do not report “no findings” until these probes have either found a concrete issue or produced evidence that the concern is handled.

Intake

Follow resources/workflow.md for the full workflow. At a glance:

  1. Identify the PR number or current branch, base branch, head branch, and whether the branch tracks a remote.
  2. Inspect the local working tree before any git operation and preserve unrelated local changes.
  3. Fetch PR metadata, commits, reviews, issue comments, changed files, mergeability, and check summary.
  4. Check whether the branch is behind its base and whether GitHub reports conflicts. Do not resolve conflicts as part of review unless explicitly asked; report conflict risk and recommend pr-to-green when needed.
  5. Check remote CI with gh pr checks or the equivalent GitHub command. Distinguish failed, pending, skipped, and missing required checks.
  6. Inspect the full PR diff against the base branch, not only the latest commit.
  7. Verify contributor claims against code, tests, and branch history before accepting them.

Classify the PR

Classify the PR by its dominant risk profile, then read the matching resource file. For mixed PRs, read every relevant resource before reviewing.

  • Feature implementation: read resources/pr-type-features.md.
  • Bug fix: read resources/pr-type-bug-fixes.md.
  • Refactor: read resources/pr-type-refactors.md.
  • Docs or notebooks: read resources/pr-type-docs-notebooks.md.
  • Data or dataset changes: read resources/pr-type-data-datasets.md.
  • Tests, CI, packaging, or infrastructure: read resources/pr-type-tests-ci-infra.md.

When classification is unclear, state the likely categories and review against the stricter applicable checklist.

Deep Dives

Read these when the PR touches the relevant surface:

  • CausalPy source-code conventions: resources/code-patterns.md.
  • Documentation and notebook conventions: resources/docs-patterns.md.
  • Severity-sorted recurring review patterns: resources/review-patterns.md.
  • Drafting or posting review comments: resources/review-comments.md.
  • Updating this skill with recurring patterns: resources/maintenance.md.
Show full SKILL.md (444 more words)Show less

Universal Checks

  • Correctness: the implementation matches the PR's stated intent, handles important edge cases, and does not introduce silent behavior changes.
  • Active probing: review comments and the PR description identify concrete concerns; independently formulate and test additional failure hypotheses that fit the changed surface. For contract, adapter, or dispatch changes, audit every producer and consumer rather than only the files named in the PR.
  • Security and privacy: no secrets, credentials, tokens, private data, unsafe deserialization, command injection, path traversal, or unnecessary network access.
  • Causal/statistical accuracy: causal claims, model assumptions, estimands, priors, simulations, and examples are technically accurate and not overstated.
  • Public API: released APIs remain compatible unless the PR intentionally changes them and documents the change; new public APIs have explicit signatures and documentation.
  • Tests: behavior changes have meaningful tests in causalpy/tests/; PyMC-heavy tests use runtime-controlled sample_kwargs; no throwaway verification scripts are added.
  • Docs: user-facing behavior changes have docs or examples where appropriate; docs follow CausalPy notebook, MyST, glossary, and citation conventions.
  • Dependencies and packaging: new dependencies are justified, declared in the right place, and do not duplicate existing tools.
  • Performance and runtime: expensive sampling, notebook execution, data loading, and CI changes are bounded and justified.
  • Maintainability: the change follows local patterns, avoids broad unrelated refactors, and keeps ownership boundaries clear.

CausalPy Review Norms

  • Before reviewing code, read AGENTS.md (workflow) and ARCHITECTURE.md (design) for core changes. For docs-heavy PRs, also inspect docs/source/notebooks/index.md; for process-sensitive PRs, inspect CONTRIBUTING.md when present.
  • Use uv run <command> (the default) for commands that import project code, run tests, build docs, or invoke repo tooling. Fall back to $CONDA_EXE run -n CausalPy <command> only when uv is unavailable. AGENTS.md and .agents/skills/python-environment/SKILL.md define both paths, including how to detect or set CONDA_EXE for the fallback.
  • Use full permissions for commands that import PyMC, PyTensor, or matplotlib to avoid false sandbox failures.
  • During review, prefer targeted local checks that match the changed surface. If you edit code or prepare a commit, run prek run during iteration and prek run --all-files before handoff unless the user explicitly says not to.
  • For markdown-only skill or docs changes, a structural read-back may be enough; report when full checks were not run and why.

Review Output

Lead with an executive summary. Explain the PR's user-facing value and the strongest evidence it delivers that value, then surface findings in severity order. If there are no findings, say which concrete probes were performed and what residual risk remains; never substitute generic praise for an evidence-backed win.

Use this structure:

markdown
## Executive Summary
Recommendation: approve / request changes / blocked / needs maintainer decision.

Value delivered:
- [Concrete user or maintainer win, tied to an implementation detail.]
- [Second concrete win when applicable.]

Review focus:
- [Highest-risk assumption and the evidence that supports or challenges it.]

## Findings

- [severity] `path`: issue, why it matters, and what should change.

## Merge Readiness
Verdict: approve / request changes / blocked / needs maintainer decision.
Branch status: up to date or behind base; conflicts if any.
CI status: green, failing, pending, skipped, or unavailable.

## PR Summary
One short paragraph describing the implementation, why it was needed, and its most important trade-off or compatibility impact.

## Test Evidence
List local and remote checks observed. Include commands only when they were actually run.

## Open Questions
Only include questions that affect merge readiness or review confidence.

When drafting comments for posting, show the draft to the user first and wait for approval. Preserve the distinction between the human maintainer's voice and any agent-authored review text.

© pymc-labs, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 12 other files in .agents/skills/review-pr of pymc-labs/CausalPy.

  • SKILL.md
  • resources/code-patterns.md
  • resources/docs-patterns.md
  • resources/maintenance.md
  • resources/pr-type-bug-fixes.md
  • resources/pr-type-data-datasets.md
  • resources/pr-type-docs-notebooks.md
  • resources/pr-type-features.md
  • resources/pr-type-refactors.md
  • resources/pr-type-tests-ci-infra.md
  • resources/review-comments.md
  • resources/review-patterns.md
  • resources/workflow.md

Open the folder on GitHubat commit 7882153

Compare with similar skills

Review PR next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Review PR compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Review PR this skillpymc-labs/CausalPy1.2k—~2.2kAutomated safety check: PassApache-2.0
Typescript React ReviewerSuFxGIT/scoutarr115—~1.7kAutomated safety check: PassNone
Cao Contributingawslabs/cli-agent-orchestrator1.4k—~4.1kAutomated safety check: PassApache-2.0
Finishing a Development Branchobra/superpowers297k5 repos~1.9kAutomated safety check: PassMIT
Typescript Advanced Typesrolling-scopes/rsschool-app10k25 repos~4.2kAutomated safety check: PassMPL-2.0
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0

Similar skills

  • Typescript React Reviewer

    SuFxGIT/scoutarr

    Expert code reviewer for TypeScript + React 19 applications.

    115 GitHub stars~1.7k tokensUpdated 6 mo ago
    DevelopmentAuto-check passed

  • Cao Contributing

    awslabs/cli-agent-orchestrator

    Official

    Contribute changes to the CAO (CLI Agent Orchestrator) codebase — the local dev loop, the CI gate map, and the pre-PR checklist.

    1.4k GitHub stars~4.1k tokensUpdated today
    DevelopmentAuto-check passed

  • Finishing a Development Branch

    obra/superpowers

    Walks the last step of a branch: confirm tests pass, detect the git environment, ask how to integrate, carry out your choice and clean up the worktree.

    297k GitHub starsUsed in 5 repos~1.9k tokens
    DevelopmentAuto-check passed

  • Typescript Advanced Types

    rolling-scopes/rsschool-app

    Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.

    10k GitHub starsUsed in 25 repos~4.2k tokens
    DevelopmentAuto-check passed

  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed

  • Check PR

    onyx-dot-app/onyx

    Checks a GitHub, GitLab, or Perforce (p4) pull request (or merge request, or shelved changelist) for unresolved review comments, failing status checks, and incomplete PR descriptions.

    32k GitHub starsUsed in 2 repos~2.3k tokens
    DevelopmentAuto-check passed

More from pymc-labs/CausalPy

All 12 skills in this repo

  • Example Datasets

    pymc-labs/CausalPy

    Load built-in CausalPy example datasets for demos, tutorials, tests, and quick causal-analysis prototypes.

    1.2k GitHub stars~587 tokensUpdated yesterday
    Auto-check passed

  • Causal Detective

    pymc-labs/CausalPy

    Challenge causal claims through structured threat assessment, counterfactual reasoning, and CausalPy falsification checks.

    1.2k GitHub stars~809 tokensUpdated yesterday
    Auto-check passed

  • Running Placebo Analysis

    pymc-labs/CausalPy

    Performs placebo-in-time sensitivity analysis with hierarchical null model and optional Bayesian assurance.

    1.2k GitHub starsUsed in 1 repo~442 tokens
    Auto-check passed

  • Python Environment

    pymc-labs/CausalPy

    Detect, configure, and use the project's Python environment (uv by default, conda-compatible tool as a fallback).

    1.2k GitHub stars~1.1k tokensUpdated yesterday
    Auto-check passed

  • Feature Exploration

    pymc-labs/CausalPy

    Explore unfamiliar APIs, libraries, or implementation behavior with minimal reproducible examples and documented findings.

    1.2k GitHub stars~442 tokensUpdated yesterday
    Auto-check passed

  • GitHub Issues

    pymc-labs/CausalPy

    Create, evaluate, and triage GitHub issues for CausalPy. An agent skill from pymc-labs/CausalPy.

    1.2k GitHub stars~254 tokensUpdated yesterday
    Auto-check passed

Categories

Questions about Review PR

What does Review PR do?

Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns. Review PR is an agent skill from pymc-labs/CausalPy. Review CausalPy pull requests end-to-end by classifying PR type, checking branch freshness, mergeability, remote CI, correctness, security, tests, docs, and maintainer concerns.

When should I use Review PR?

Review PR fits situations like: asked to review a PR; assess a branch before merge; summarize PR risks; request changes.

How do I install Review PR in Claude Code?

Run `npx skills add pymc-labs/CausalPy --skill review-pr -a claude-code`. Or copy the skill folder (.agents/skills/review-pr in pymc-labs/CausalPy) into .claude/skills/review-pr in your project. Claude Code loads it when a task matches its description.

How do I install Review PR in Codex?

Run `npx skills add pymc-labs/CausalPy --skill review-pr -a codex`. Or copy the skill folder (.agents/skills/review-pr in pymc-labs/CausalPy) into .agents/skills/review-pr in your project. Codex loads it when a task matches its description.

Can I use Review PR in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add pymc-labs/CausalPy --skill review-pr -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/review-pr, .gemini/skills/review-pr, .github/skills/review-pr and .opencode/skills/review-pr in your project.

What does Review PR need to run?

Going by SKILL.md and its folder, Review PR needs the command-line tools its instructions call (gh and uv).

Does Review PR access the network?

SKILL.md contains no URLs. Its commands use gh and uv, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Review PR safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Review PR use?

Review PR is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Review PR use?

About 2.2k tokens (SKILL.md is roughly 8.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Review PR?

Skills that share tags, products or a category with Review PR: Typescript React Reviewer (SuFxGIT/scoutarr, 115 stars), Cao Contributing (awslabs/cli-agent-orchestrator, 1.4k stars), Finishing a Development Branch (obra/superpowers, 297k stars) and Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Review PR?

pymc-labs (a GitHub organization) maintains it in pymc-labs/CausalPy, which has 1,201 GitHub stars. The repository holds 12 skills in this directory. The repository was last updated on October 9, 2026.

Source: pymc-labs/CausalPy on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.