dayan-agent-skills

Public playbooks

Playbooks turn the method cards into repeatable operating routes. Use them when a Skill alone is too small and a full agent system is too much.

Available playbooks

• Control-layer design: decide whether a repeatable AI workflow should be a checklist, Skill, hook, verifier, or agent.
• From conversation to Skill: convert repeated working behavior into a clean public Skill package.
• Release review loop: run a separated builder/checker path before publishing an AI-assisted artifact.

How to use a playbook

1. Pick one concrete repeated problem.
2. Choose the smallest control layer that can hold the problem.
3. Produce one accepted example and one rejected or boundary example.
4. Add the evidence that would make the result stronger.
5. Keep the human release authority explicit.

The playbooks are public methods, not claims that every scenario has already been externally validated.

This site is open source. Improve this page.