pi¶
Install pi and add the stax skill so pi can drive stax workflows correctly.
1. Install¶
See pi.dev for installation.
2. Add the stax skill¶
The easiest path is to let stax manage it:
stax skills update
This writes the skill to ~/.pi/agent/skills/stax/SKILL.md (alongside the other agents). To do it manually:
mkdir -p ~/.pi/agent/skills/stax
st --skill > ~/.pi/agent/skills/stax/SKILL.md
# or fetch the markdown body only from GitHub:
curl -o ~/.pi/agent/skills/stax/SKILL.md https://raw.githubusercontent.com/cesarferreira/stax/main/skills.md
pi loads skills from ~/.pi/agent/skills/<name>/SKILL.md.
3. Use pi with AI create/PR generation¶
st generate --pr-body --agent pi
st generate --pr-body --agent pi --model anthropic/claude-opus-4-8
st gen --pr-title --agent pi
st gen --commit-msg --agent pi
4. AI worktree lanes¶
st lane deep-dive --agent pi
st lane deep-dive --agent pi --model anthropic/claude-opus-4-8 "trace the flaky test"
--yolo is not supported for pi (permission bypass is provided by opt-in
permission-mode extensions, not a stable core CLI flag). Pass a bypass flag
manually with --agent-arg if your pi setup supports one.