PI.md

Pi Adapter

Read AGENTS.md first - it carries the operating rules, the required reads, the session lifecycle, and the skill routing for every tool. This file is only what differs in Pi. Pi is a minimal harness: the loop's behaviour comes entirely from these files, not from anything Pi does on its own.

loop session-start --tool pi --command "<slash-command>"
loop session-end --tool pi

Two namespaces

Pi reserves /skill: for skills and leaves plain / to prompt templates, so loop skills writes both:

	Path	Reached by
Skills	~/.pi/agent/skills/loop-<name>/SKILL.md	The model from the description, or /skill:loop-<name>
Prompt templates	~/.pi/agent/prompts/<name>.md	The user typing /<name>

Installing only skills leaves /plan-loop matching nothing. Verify with /help in Pi, or:

ls ~/.pi/agent/skills ~/.pi/agent/prompts

Pi also scans ~/.agents/skills globally and .agents/skills in the cwd and its ancestors, so loop skills install --project reaches Pi through the universal directory - no .pi/skills copy, which Pi would list twice.

Project scope needs trust. Pi reads .agents/skills, .pi/skills, and .pi/prompts only after the project is trusted (~/.pi/agent/trust.json). Trust the folder once; a global install needs none of this.

Instructions files

Pi loads AGENTS.md from ~/.pi/agent/, the cwd, and its ancestors, so a workspace containing this repo's AGENTS.md needs no configuration. SYSTEM.md can override the system prompt per project; Loop writes none, and a project that adds one must keep AGENTS.md routing intact.