| name | skill-creator |
| description | Save successful workflows as reusable pi skills. Use when a workflow worked well and should be preserved for future use. |
Create Skills from Experience
When to Use
After completing a task that involved a non-obvious workflow:
- Multi-step debugging process that worked
- Build/deploy procedure that required specific steps
- Integration pattern that took multiple attempts to get right
- Any workflow where you think "I should remember how to do this"
How to Create a Skill
- Identify the workflow โ what steps did you follow? What worked?
- Write the SKILL.md โ save to
~/.agents/skills/<name>/SKILL.md
- Keep it specific and actionable โ steps, commands, verification
SKILL.md Format
---
name: skill-name
description: "One-line description of what this skill does and when to use it."
---
# Skill Title
## When to Use
- Trigger condition 1
- Trigger condition 2
## Steps
1. Step one with exact commands
2. Step two
3. Verification step
## Pitfalls
- Common mistake and how to avoid it
Rules
- name: lowercase, hyphenated, max 64 chars
- description: one line, max 1024 chars, explains WHEN to use it
- Body: concrete steps, exact commands, verification
- No fluff: skip background, motivation, history โ just the procedure
- Include pitfalls: what goes wrong and how to avoid it
- Include verification: how to confirm it worked
Where to Save
mkdir -p ~/.agents/skills/<skill-name>
Skills in ~/.agents/skills/ are user-local and automatically discovered by pi.
Good vs Bad Skills
Good skill: Specific procedure with exact commands
## Steps
1. Run `uv run pytest tests/ -q` to get baseline
2. Check for `FAILED` in output
3. If failures, run `uv run pytest tests/test_specific.py -v` for details
Bad skill: Vague advice
## Steps
1. Run tests
2. Check results
3. Fix issues