Install pi and add the stax skill so pi can drive stax workflows correctly.
See pi.dev for installation.
The easiest path is to let stax manage it:
stax skills updateThis 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.mdpi loads skills from ~/.pi/agent/skills/<name>/SKILL.md.
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 pist 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.