# Agentic Workflows

> Design, create, update, debug, audit, or upgrade GitHub Agentic Workflows with the `gh aw` extension. Use when work involves workflow Markdown, compiled `.lock.yml` files, agent engines, MCP tools, safe outputs, or `gh aw` commands.

- Skill: `timmo001/agentic-workflows` (Agent Skill)
- Install (CLI): `npx skillmds@latest add timmo001/agentic-workflows`
- Raw SKILL.md: https://api.skillmd.com/api/skills/timmo001/agentic-workflows/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: timmo001 (https://skillmd.com/u/timmo001)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/timmo001/agentic-workflows

---


# Agentic Workflows

Use GitHub's `gh aw` extension for agentic workflows written as Markdown and
compiled to GitHub Actions lock files.

## Workflow

1. Read the workflow Markdown, its compiled `.lock.yml`, and
   `.github/aw/instructions.md` when present. Repository instructions override
   general guidance.
2. Check the installed interface with `gh aw --help` and the relevant
   subcommand help.
3. Read only the relevant current documentation or prompt from the official
   [`github/gh-aw`](https://github.com/github/gh-aw) repository. Use its
   `.github/aw/` guidance for the task type instead of guessing syntax:
   - design: `designer.md`
   - create: `create-agentic-workflow.md`
   - update: `update-agentic-workflow.md`
   - optimise: `optimize-agentic-workflow.md`
   - debug or audit: `debug-agentic-workflow.md`
   - upgrade: `upgrade-agentic-workflows.md`
   - engine setup: `configure-agentic-engine.md`
   - agent runtime, Docker, gVisor, Docker sbx, ARC DinD, self-hosted
     runners, or runtime installation: `agent-runtime-instructions.md`
   - shared workflows or MCP wrappers: `create-shared-agentic-workflow.md`
   - drive-backed workflow memory: `drive-memory.md`
   - declarative skills or agent plugins: `skills.md`
   - report generation: `report.md`
   - Dependabot manifest updates: `dependabot.md`
   - test coverage analysis: `test-coverage.md`
   - compact Markdown charts: `asciicharts.md`
   - CLI command to MCP tool mapping: `cli-commands.md`
   - workflow architecture and patterns: `patterns.md`
   - token usage and cost optimisation: `token-optimization.md`
   - long-running multi-agent research: `multi-agent-research.md`
   - OpenTelemetry queries and analysis: `skills/otel-queries/SKILL.md`
   - permissions and output controls: `safe-outputs.md` and
     `workflow-constraints.md`
4. Edit the Markdown source, not the generated lock file. Compile with the
   repository's documented command and review the generated diff.
5. Run the narrowest available validation or audit command for the workflow.

## Safety

- Treat workflow permissions, credentials, network access, imported tools,
  MCP servers, and safe outputs as security boundaries.
- Keep permissions and network access at the minimum required for the task.
- Prefer safe-output mechanisms over direct write-capable GitHub tools.
- Never hand-edit generated `.lock.yml` files.
- Do not run, enable, push, or dispatch a workflow without the user's explicit
  authorisation for that action.
- Verify current syntax against the installed extension and official source.
  `gh-aw` evolves quickly, so do not rely on remembered flags.