# github/gh-aw

GitHub Agentic Workflows

Repository: https://github.com/github/gh-aw
Canonical: https://ross.abutalabs.com/products/gh-aw
Homepage: https://gh.io/gh-aw
Language: Go
License: MIT
License Family: permissive
Topics: gh-extension, actions, cai, ci, claude-code, codex, copilot, github-actions
Last push: 2026-08-26T23:24:17+00:00

## Health v2 (maintenance only)
Score: 80/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 85, longevity 27
- inputs: {"age_days": 386, "days_push": 7, "days_rel": 22, "gap_med": 0.0, "n_releases_24m": 349}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5016, forks 507 (observed 2026-08-28T04:09:04.647012+00:00)

## What it is
GitHub Agentic Workflows (gh-aw) is a GitHub CLI extension that lets developers define AI-powered repository automation in Markdown with YAML frontmatter, compiling it into standard GitHub Actions workflows. It runs AI agents like Copilot, Claude Code, Codex, and Gemini securely in CI, with read-only sandboxed defaults and validated safe-outputs for writes.

## Use cases
- automate issue triage with an AI agent in GitHub Actions
- run AI-powered pull request review on every PR
- investigate CI failures automatically with an agent
- keep documentation up to date using AI workflows
- define AI repository automation as Markdown files
- analyze dependencies and generate repository reports with agents

## When to choose
- you want AI agents to handle tasks needing reasoning like triage, review, or CI investigation
- you want agentic automation with strong security defaults: sandboxing, read-only agents, scoped safe outputs
- you already use GitHub Actions and want to complement CI with AI workflows
- you want to choose among Copilot, Claude Code, Codex, Gemini, or Pi as the agent engine

## When to avoid
- you only need deterministic builds, tests, linting, or deployments - plain GitHub Actions suffices
- you need fully deterministic, reproducible automation with no AI variability
- you don't use GitHub Actions or the GitHub ecosystem
- you can't accept the cost or billing implications of running LLM agents in CI

## Facets
- artifact type: cli-tool
- maturity: active
- function: ci-cd, workflow-automation, agent-framework, developer-tools, cli
- domain: developer-tools, large-language-models
- platform: cli, go, cloud
- tags: github-actions, gh-extension, agentic-workflows, claude-code, copilot, codex, issue-triage, pr-review, markdown-workflows, sandboxed-agents, automation, devops, ai-agents, docker

## Member repositories
- github/gh-aw (main) score 80

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:04.647012+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:17:31.650574+00:00, confidence not recorded.
  - readme: https://github.com/github/gh-aw (fetched 2026-08-28T04:09:04.647012+00:00, sha d5043a8e71d8)
- Data as of 2026-08-30T08:39:29.467469+00:00.