pi-config

Initialize your project configuration, spin up the background daemon, and run your first custom agent workflow in 60 seconds.

Configure the automated code review loop to use Qodo and CodeRabbit for continuous pull request feedback.

Create, route, and test new specialist agents to handle specific tasks like security auditing or test running.

Monitor background tasks, inspect code diffs, and switch between sessions using the local React UI.

Seed, score, and organize learned topics to give your agents persistent context across sessions.

Use /btw to ask a quick model question using conversation context without adding the exchange to the main session history.

Register custom chat commands to quickly trigger specific AI scripts or automations.

Add enforcement rules to intercept and block destructive bash commands or sensitive git operations.

Complete command index for database queries, review handling, and manual memory management.

Comprehensive lookup for pi project settings, UI configuration, and environment variables.

Catalog of shipped prompt commands (/implement, /implement-and-review, /issue-review, /pr-review, /review-local, /refine-review, /release, /remember, /create-skill, /domain-model, /query-db): arguments, prerequisites, and when to use each.

Understand the underlying vector embeddings, topic scoring algorithms, and database migrations.

Deep dive into the background task queues, async agent state handling, and inter-process communication.

How to use Pi within Neovim, populate the quickfix list with git changed files, and leverage the remote lua execution capabilities.

Setting up the background Discord bot to broadcast daemon events, review handler updates, and agent status notifications.

Configuring and using the agent image generation tool with Gemini, including supported aspect ratios and output handling.

Install and configure the google-vertex-claude provider, authenticate with gcloud, select models (including 1M context variants via VERTEX_CLAUDE_1M), and run pi with --provider google-vertex-claude.

Run the pi-sidecar server, manage AI sessions over REST, discover models from ACPX and CLI providers, configure custom HTTP-backed tools, and integrate with the Python SidecarClient.

Using the External AI CLI to interact with Cursor, Claude, and Gemini from the terminal, including model configuration and prompting.

How to spawn non-blocking async agents, register cron schedules for recurring workflows, and monitor background tasks via the UI.

Explains how multiple agent sessions communicate, broadcast messages, and synchronize directories using the local communications server.

Details the automated background worker that runs periodically to extract and consolidate project memories, skills, and enforcements from recent sessions.

Guide to the ACPX runtime modules, explaining how external agent models are discovered and executed natively.