REPO PACKAGE
github gh-aw
OFFICIAL42skills
22ready to use
LABSCO SUMMARY
The product gh-aw compiles agentic workflows written in markdown into GitHub Actions, and exactly one skill, named gh-aw, actually explains that to a user. The other 42 sit closer to the repository's own Go codebase: go-linters and go-codemod validate and generate the project's custom static-analysis rules, checkout-credential-review and http_mcp_headers review credential and header handling in gh-aw's own code, error-messages and error-pattern-safety standardize how gh-aw itself reports a failure, and pr-to-go-linter turns a merged pull request into a new linter for the next one.
A smaller group is genuinely reusable outside this repository: github-issue-query, github-pr-query, github-discussion-query, and github-labels-query are jq-filtered wrappers over gh's own query commands, github-script and playwright-cli are general scripting and browser-automation references, and prompt-token-efficiency and reporting are generic writing guidance with nothing gh-aw-specific in them. Someone who wants to automate GitHub from an agent, rather than contribute to gh-aw, will get more use out of this handful than out of the rest of the 43.
READ THE FULL ANALYSIS
Two of the forty-three are unexplained. sergo-examples ('Optional Sergo examples for cache formats and reporting templates') and skillz-integration ('Run and integrate Skillz MCP server with Docker for skill execution') both name a system — Sergo, Skillz — that is never defined in the README or in any of the other skill descriptions, so it is not possible to say from this material what either one actually does.
The README carries an active warning worth repeating here. As of the version we captured, gh-aw releases 0.68.4 through 0.71.3 are being retired over a bug that affects billing, and anyone on one of those versions is told to upgrade immediately. That notice is about the CLI tool rather than the skills, but it is the kind of thing a reader landing on this page for the skills would otherwise miss.
Of the 43, 24 need nothing beyond installing them, 10 want a local tool already on the machine, 3 want an account key on file, 3 talk to an MCP server, and 3 the site's own checker could not classify at all.
ALSO IN THIS PACKAGE
AGENT PERSONALITIES
12 itemsAGENT PERSONALITY
adr-writer
Best-practice Architecture Decision Record (ADR) writer following the Michael Nygard template. Generates, revises, and stores ADRs in docs/adr/.
AGENT PERSONALITY
Agentic Workflows
GitHub Agentic Workflows (gh-aw) - Create, debug, and upgrade AI-powered workflows with intelligent prompt routing.
AGENT PERSONALITY
ci cleaner
Tidies up the repository CI state by formatting sources, running linters, fixing issues, running tests, and recompiling workflows
AGENT PERSONALITY
contribution checker
Evaluate a single PR against the target repository's CONTRIBUTING.md for compliance and quality
AGENT PERSONALITY
create safe output type
Adding a New Safe Output Type to GitHub Agentic Workflows
AGENT PERSONALITY
custom engine implementation
Comprehensive guide for implementing custom agentic engines in gh-aw
AGENT PERSONALITY
developer.instructions
Developer Instructions for GitHub Agentic Workflows
AGENT PERSONALITY
grumpy-reviewer
Critical code reviewer with 40+ years of experience who performs thorough reviews with a grumpy, sarcastic tone
AGENT PERSONALITY
interactive agent designer
Interactive wizard that guides users through creating and optimizing high-quality prompts, agent instructions, and workflow descriptions for GitHub Agentic Workflows
AGENT PERSONALITY
Squad
Your AI team. Describe what you're building, get a team of specialists that live in your repo.
AGENT PERSONALITY
technical-doc-writer
AI technical documentation writer for GitHub Actions library using GitHub Docs voice
AGENT PERSONALITY
w3c-specification-writer
AI technical specification writer following W3C conventions and best practices for formal specifications
PROMPT TEMPLATES
5 itemsPROMPT TEMPLATE
cache_memory_prompt
PROMPT TEMPLATE
cache_memory_prompt_multi
PROMPT TEMPLATE
checkouts_no_credentials_warning
PROMPT TEMPLATE
github_context_prompt
PROMPT TEMPLATE
threat_detection
MCP SERVERS
2 itemsMCP SERVER
github-agentic-workflows
stdio · gh aw mcp-serverMCP SERVER
serena
Serena MCP server for semantic code retrieval, code editing, and IDE-like context tools that help AI coding agents work inside a codebase.
stdio · uvx --from git+https://github.com/oraios/serena serena start-mcp-server --context codex --project ${workspaceFolder}WHAT'S INSIDE
42 showing · 42 totalReady to use
Nothing else to set up — install it and go.
42 of 42
SETUP
Showing all 42 skills
agentic-workflows
Points a request about GitHub's AI-run repository automations at the one reference page that covers it, instead of loading dozens.
awf-release-integrator
A maintainer's routine for moving the bundled firewall inside GitHub's agentic-workflow tool up to its newest release, and for spotting what else that release changes.
checkout-credential-review
A review of code that clones a repository and then runs git commands, checking whether an access token is actually available at that point and whether the code copes when, by design, there is none.
console-rendering
Takes the layout work out of printing Go data in a terminal: you tag each field with how it should appear, and the output comes back aligned.
copilot-review
A method for working through pull request feedback without dropping anything, from the first comment you read to the last thread you close.
create-canvas
Gives a Copilot conversation something visual to show: a small web page the assistant can open in a side panel, hand data to, and act on when you click.
custom-agents
How you write down the rules GitHub Copilot should follow in your repository, and where the file has to live for Copilot to pick it up.
debugging-workflows
A troubleshooting manual for GitHub workflows that hand the job to an AI agent — what to look at when a run fails, and what the usual causes turn out to be.
developer
The step-by-step build-and-check routine a contributor follows when changing the GitHub Agentic Workflows codebase — which command to run after each kind of edit, when to run the full test suite, and how to merge main back in cleanly.
documentation
Writing standards for the GitHub Agentic Workflows documentation site: what tone to use, what shape a page takes, and where to stop.
error-messages
The rule this project applies to its own error text: never tell someone that something is wrong without showing them what right looks like.
error-pattern-safety
Stops a badly written search pattern from freezing the program: a pattern that can match nothing at all will scan the same spot forever, and these rules keep that out of the code.
error-recovery-patterns
Draws the line between a temporary hiccup and a failure that will never fix itself, and says what the code should do in each case.
gh-agent-session
The command-line view of the coding jobs Copilot is already running on a repository — what is open, how far along each one is, and what came out of it.
gh-agent-task
A way to hand a coding job to GitHub Copilot without leaving the terminal: you write the instructions, it opens an issue, does the work, and comes back with a pull request.
gh-aw
The starting point for gh-aw, GitHub's add-on that takes a task written as an ordinary markdown file and runs it automatically on GitHub.
github-copilot-agent-tips-and-tricks
GitHub's Copilot opens pull requests of its own, and they get lost among everyone else's — this is how to find them, narrow the list down, and see what each one changed.
github-discussion-query
Gets the discussion threads out of a GitHub repository as plain data, so you can count and sort them yourself instead of scrolling the page.
github-issue-query
Answers questions about a repository's issues as data you can feed into something else, rather than as a page you have to read and count by hand.
github-labels-query
Lists the labels a repository uses — the small tags like bug or enhancement that get attached to its issues and pull requests.
github-mcp-server
Sets up the bridge that lets an AI assistant act on GitHub itself — reading issues, opening pull requests, checking builds — and how much of that it may do.
github-pr-query
Pulls a repository's pull requests into a list you can narrow to just the ones you need — the open ones, one person's, or the ones still waiting on a review.
github-script
How to write the little JavaScript snippets that run inside a GitHub Actions step so they log, set outputs, and report failure the right way — and the easy mistakes that silently break them.
github-workflows-query
The automated jobs a repository runs on GitHub, listed with their names, their files and whether each one is switched on.
go-codemod
Writing one of the automatic fixers that brings an out-of-date gh-aw workflow file up to the current format — and the tests it must pass first.
go-linters
Add your own automatic check to this project's Go code, from the package it lives in to the line that turns it on.
http_mcp_headers
Passing an API key to an outside service a workflow talks to, without the key itself ever being written into the generated config file.
javascript-refactoring
Splitting a big JavaScript file into small pieces that each have their own tests, and doing it the way this repository expects.
jqschema
Turns a wall of JSON into a short outline of its structure, so an unfamiliar response is readable at a glance.
messages
Adding a new message gh-aw can send, one whose wording a workflow author can override, and registering it everywhere the build expects to find it.
optimize-agentic-workflow
Works out why an automated AI job on GitHub is costing so much, and gives you a ranked plan for making it cheaper.
otel-queries
Reading the records a workflow run leaves behind — timings, errors, gaps — until one concrete cause is proved rather than guessed.
playwright-cli
You drive a real browser step by step from the command line, getting a snapshot of the page back after every action.
pr-finisher
Takes an open pull request the whole way to mergeable: review comments answered, tests passing locally, conflicts cleared.
pr-to-go-linter
Turns the mistake one pull request fixed into an automatic check, so the same mistake gets caught everywhere else in the codebase.
prompt-token-efficiency
Writing to an AI the way it actually reads best — short, one word per idea, and a plain statement of what a finished answer looks like.
reporting
Stops a long report from burying its reader in scroll: it goes out folded shut, and opens only where someone wants the detail.
sergo-examples
Ready-made format examples to copy when Sergo has to write down its tools, its history, a task or a report.
skillz-integration
Hands a folder of your own written-down procedures to the AI running an automated GitHub job, so it can pick them up and follow them mid-run.
ssl-skill-normalizer
Turns a skill written as prose into a structured file a machine can search and check: when it runs, what stages it has, and what each step does.
temporary-id-safe-output
Lets an agent refer to an issue it has not created yet, using a placeholder that is swapped for the real issue number once the issue exists.
workflow-step-summaries
How to write the little report a build step leaves on its run page, so the outcome is at the top and the long logs stay folded away.
No skill in this package matches that.
HOW TO GET IT
WHOLE PACKAGEInstall the whole package (42 skills)
npx skills add github/gh-awSINGLE SKILLOr install a single skill
npx skills add github/gh-aw --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.