RE

review-project-pr

Perform local-only PR reviews for Brainhack School projects, including build tests and metadata checks.

Install

mkdir -p .claude/skills/review-project-pr && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/19081" && unzip -o skill.zip -d .claude/skills/review-project-pr && rm skill.zip

Installs to .claude/skills/review-project-pr

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

This skill should be used when the user asks to review a Brainhack School pull request by number (e.g., "/review-project-pr 437" or "review PR 437"). It checks that the PR builds with Hugo, that images are not too heavy, that no stray files (notebooks, data) were added, that the frontmatter date is credible, that links work, and that the content has no typos or leftover garbage. It produces a review report in chat plus a local hugo serve link to inspect the rendered project page. Local-only — it never posts to GitHub.
523 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Intermediate

Key capabilities

  • Fetch a pull request into an isolated worktree
  • Run mechanical checks on the PR, including Hugo build validation
  • Check for image optimization and stray files
  • Perform qualitative review of frontmatter and content
  • Serve a local preview of the rendered project page
  • Generate a review report for PR feedback

How it works

This skill fetches a specified pull request into an isolated worktree, runs a series of mechanical and qualitative checks, and then serves a local preview of the project page.

Inputs & outputs

You give it
Pull request number
You get back
Review report in chat and a local Hugo serve link

When to use review-project-pr

  • Reviewing student project PRs
  • Validating Hugo site builds
  • Checking project metadata

About this skill

Review a Brainhack School project PR

Review a student-project PR against the site's conventions and produce a report the user can copy into PR feedback. Everything stays local: never comment on the PR, never push, never modify the main working tree.

Environment facts

  • Main repo: /home/pbellec/git/school-brainhack.github.io (remote origin = the upstream school-brainhack repo).
  • Use the pinned Hugo at ~/git/brainhack-pr-review/bin/hugo (extended 0.128.0, same version as CI; run_checks.sh auto-downloads it). Do NOT use /usr/bin/hugo (0.92, too old) or /snap/bin/hugo (0.164, rejects a duplicate YAML key in config.yaml that CI tolerates).
  • Worktrees do not contain the theme submodule; always pass --themesDir /home/pbellec/git/school-brainhack.github.io/themes to hugo.

Workflow

1. Fetch the PR into an isolated worktree

bash .claude/skills/review-project-pr/scripts/setup_pr_worktree.sh <PR#>

This creates ~/git/brainhack-pr-review/pr-<n> and prints the files changed vs origin/main. Also grab metadata:

gh pr view <PR#> --json title,author,createdAt,body,url

2. Run the mechanical checks

bash .claude/skills/review-project-pr/scripts/run_checks.sh ~/git/brainhack-pr-review/pr-<n>

This prints [BUILD], [IMAGES], [STRAY], and [LINKS] sections with PASS/WARN/FAIL lines. Interpret them; downgrade link FAILs to WARN when the site is known to block bots.

3. Qualitative review

Read references/checklist.md for the full rubric, then read the PR's index.md in the worktree and check:

  • Date credibility: frontmatter date: vs the PR's createdAt and the current school edition.
  • Frontmatter conventions: type: "project", lowercase tags, ~75-word summary, image: file exists, hyphenated slug.
  • Internal links/images: relative references in index.md resolve, and the built page under <worktree>/public/project/<slug>/ contains them.
  • Typos: genuine misspellings/grammar errors with locations — not style nits.
  • Garbage: leftover template text, duplicated blocks, broken shortcodes, merge-conflict markers, secrets, unrelated content.

4. Serve the preview

Kill any previous preview, then start hugo serve in the background on port 1414 (1313 is left free for the user's own serve):

pkill -f "hugo serve.*1414" 2>/dev/null
cd ~/git/brainhack-pr-review/pr-<n> && ~/git/brainhack-pr-review/bin/hugo serve --port 1414 --bind 127.0.0.1 \
  --themesDir /home/pbellec/git/school-brainhack.github.io/themes

Run it with run_in_background: true. Determine the slug's rendered URL path (Hugo lowercases it) and include http://localhost:1414/project/<slug>/ in the report.

5. Report

Use the report template at the end of references/checklist.md: lead with an overall verdict (✅ / ⚠️ / ❌), then a summary table, then findings written so they can be copy-pasted as PR feedback, then the preview link.

Cleanup

Reviewing the same PR again refreshes the same worktree automatically. To clean up manually: git worktree remove --force ~/git/brainhack-pr-review/pr-<n> from the main repo.

When not to use it

  • When the user wants to post comments directly to GitHub
  • When the user wants to push changes to the main repository
  • When the user wants to modify the main working tree

Limitations

  • The skill never posts to GitHub
  • The skill never pushes changes
  • The skill never modifies the main working tree

How it compares

This workflow provides a structured, local-only review process for pull requests, ensuring privacy and consistency, unlike directly interacting with GitHub or performing manual checks.

Compared to similar skills

review-project-pr side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
review-project-pr (this skill)012dReviewIntermediate
deepwiki-rs259moReviewIntermediate
python-code-style95moReviewIntermediate
code-review-excellence194moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

deepwiki-rs

sopaco

AI-powered Rust documentation generation engine for comprehensive codebase analysis, C4 architecture diagrams, and automated technical documentation. Use when Claude needs to analyze source code, understand software architecture, generate technical specs, or create professional documentation from any programming language.

25170

python-code-style

wshobson

Python code style, linting, formatting, naming conventions, and documentation standards. Use when writing new code, reviewing style, configuring linters, writing docstrings, or establishing project standards.

971

code-review-excellence

wshobson

Master effective code review practices to provide constructive feedback, catch bugs early, and foster knowledge sharing while maintaining team morale. Use when reviewing pull requests, establishing review standards, or mentoring developers.

1958

code-walk-thru

pchalasani

Use this when user wants you to walk through (code or text) files in a EDITOR to either explain how some code works, or to show the user what changes you made, etc. You would typically use this repeatedly to show the user your changes or code files one by one, sometimes with specific line-numbers. This way the user is easily able to follow along in their favorite EDITOR as you point at various files possibly at specific line numbers within those files.

670

cookbook-audit

anthropics

Audit an Anthropic Cookbook notebook based on a rubric. Use whenever a notebook review or audit is requested.

568

schema-markup

davila7

When the user wants to add, fix, or optimize schema markup and structured data on their site. Also use when the user mentions "schema markup," "structured data," "JSON-LD," "rich snippets," "schema.org," "FAQ schema," "product schema," "review schema," or "breadcrumb schema." For broader SEO issues, see seo-audit.

1042

Search skills

Search the agent skills registry