Getting Started

Install the P31 CLI, configure the router proxy, and run your first chat.

Prerequisites

Linux (x86-64 or arm64) or macOS (Intel or Apple Silicon)
Node.js 18+ (for MCP servers)
A DeepSeek API key or other OpenAI-compatible endpoint

Installation

One-line install:

curl -fsSL https://p31-cli.pages.dev/install | bash

Or via npm:

npm install -g @p31/cli

Verify the installation:

p31 version
p31 help

Set Up the Router Proxy

The router proxy (:4001) forwards AI requests to your model provider. Start it alongside the CLI:

p31 router start

Verify it's running:

curl http://localhost:4001/health
# Expected: {"status":"ok"}

Configuration

The CLI reads from ~/.p31/config.yaml:

proxy_url: http://localhost:4001/v1
proxy_model: flash
ollama_url: http://localhost:11434

Edit this file to change your model (flash, premium, scavenger) or proxy URL.

Your First Chat

Start the interactive chat TUI:

p31 chat

Type a message and press Enter. The agent will respond using the router proxy.

Press Ctrl+P to open the command palette with 28+ commands. Press Esc or Ctrl+C to quit.

MCP Servers: The chat automatically connects to PHOS Forge (26 tools) and Google Workspace on startup. You'll see connection status in the chat window.

Next Steps

Learn the CLI commands
Explore MCP server integration
Understand the Spatial Oasis mesh