This repository produces artifacts that can be used by both humans and AI coding agents.
Use skills when you want an assistant to repeatedly apply the same engineering judgment.
Typical use:
- Copy the skill directory into your assistant's skill folder.
- Keep the
SKILL.mdfrontmatter intact. - Invoke the skill by name or by asking for the behavior described in its
description.
Good for:
- Code review workflows
- Performance review
- Teaching sessions
- AI coding guardrails
Use prompts when you want a portable instruction that works in many chat or coding tools.
Good for:
- One-off implementation requests
- Asking another agent to stay grounded
- Reviewing generated code
- Creating consistent team prompts
Use checklists when a human reviewer needs a compact decision aid.
Good for:
- Pull request review
- Architecture review
- Pairing sessions
- Team working agreements
Use workflows when the process matters more than a single instruction.
Good for:
- Adapting a new source
- Teaching the repository method
- Creating future cases
For the first case:
- Read Case 001.
- Try llm-code-grounding-loop in a real coding task.
- Use agentic-grounding-checklist.md to review the output.