Agent Workflow Pack

Reusable workflow packs for AI coding agents.

Landing page: https://memoash.github.io/agent-workflow-pack/

Free fit check: https://github.com/MemoAsh/agent-workflow-pack/issues/new?template=pilot.yml

Example fit check submission: #1

Free SKILL.md checker: https://memoash.github.io/agent-workflow-pack/skill-checker.html

Try PR review without local setup: https://memoash.github.io/agent-workflow-pack/try-pr-review.html

Claude Code users can copy a workflow directory into .claude/skills/. Codex users can copy the same workflow into .codex/skills/. Keep repo-wide rules in AGENTS.md; keep repeated task procedures in SKILL.md.

The top-level skills/ directory is intentionally indexer-friendly for skill directories and marketplace crawlers.

Quick install examples:

# Claude Code
cp examples/pr-review-workflow/AGENTS.md ./AGENTS.md
mkdir -p .claude/skills/pr-review
cp -R skills/pr-review-workflow/* .claude/skills/pr-review/

# Codex
cp examples/pr-review-workflow/AGENTS.md ./AGENTS.md
mkdir -p .codex/skills/pr-review
cp -R skills/pr-review-workflow/* .codex/skills/pr-review/

If you keep pasting the same long prompt into Codex, Claude Code, Copilot, Cursor, or another coding agent, that workflow probably belongs in a reusable pack:

• durable repo instructions
• task-specific skill/checklist
• optional output templates
• verification steps

This turns repeated prompt knowledge into something your future self or team can run again.

The Problem

AI coding agents are powerful, but repeated workflows often stay trapped in chat:

• "Review this PR, but only report real risks."
• "Fix this bug, but reproduce it first."
• "Write release notes in our format."
• "Never edit generated files."
• "Run these checks before saying done."

When those rules live only in a prompt, they drift. People rewrite them, agents forget them, and teams cannot audit what actually happened.

The Pack

A workflow pack gives the agent a repeatable operating procedure.

Typical layout:

repo/
  AGENTS.md
  skills/
    workflow-name/
      SKILL.md
      verification.md
      templates/
        output-template.md

What Goes Where

Need	Put It Here
Repo-wide conventions	AGENTS.md
Task procedure	skills/<workflow>/SKILL.md
Expected output format	templates/
Success checks	verification.md
Mechanical repeatable steps	scripts/ when needed

Example

See examples/pr-review-workflow for a small PR-review workflow pack.

More examples:

• Installable PR review skill
• PR review workflow
• Safe edit workflow

Service

I offer a small paid pilot:

• $99: one narrow workflow pack
• $299: one workflow pack, setup guidance, and one revision after testing

Good pilot workflows:

• PR review
• bug reproduction and fix
• release notes
• safe edit / restricted path checks
• QA verification
• customer issue triage
• internal handoff for weaker AI models

Not a good fit:

• spam, account abuse, platform evasion, or illegal scraping
• vague "make my agent better" requests
• full app builds disguised as one workflow
• generic prompt packs with no verification

How A Pilot Works

1. You send one repeated prompt, SOP, checklist, or example output.
2. I separate repo rules, task steps, templates, and verification.
3. I deliver a small workflow pack.
4. You test it once.
5. I make one revision based on what failed.

Contact

Send one repeated AI-agent workflow you are tired of rewriting.

If it is a bad fit, I will say that before you pay.

Open a free fit check / pilot issue with your repeated workflow.

See service/offer.md for scope and pricing.

License

MIT. See LICENSE.