Introduction

What is Pasture Protocol?

Pasture Protocol is a local AI companion runtime. It installs the `pasture` CLI, runs a daemon on your computer, connects to WhatsApp and Telegram, serves a local dashboard, and uses whichever local or cloud LLM providers you configure. There is no Pasture-hosted message broker routing your chats.

What these docs explain

These docs are written around the actual runtime: prompts and skill Markdown define agent behavior, JavaScript routes messages and executes tools, and local state under `~/.pasture` holds configuration, workspace files, memory, reminders, logs, and channel auth. Start with use cases if you want examples; start with request handling if you want the system model.

How to read these docs

We start simple: concepts first, implementation later. The Agents section explains named personas and delegation. The request lifecycle page explains the router, intent planning, skill loading, and confirmation boundaries. Deeper pages cover skills, missions, channels, reports, and privacy.

Docs map

• Use cases: what you can ask Pasture to do
• How requests are handled: routing, context, skills, approval, reply
• Channels: dashboard, WhatsApp, Telegram, groups
• Agents overview: team, personas, main
• Delegation: agent-send, coordinators
• Skills: tools per agent
• Missions: persistent work
• Routing: DMs and groups
• Reports: agent throughput, delegation, skill usage
• Privacy & safety: local runtime, secrets, confirmations

Install & runtime

Install with the one-line script from the hero, then use `pasture start`, `pasture dashboard`, `pasture auth`, `pasture logs`, and `pasture update`. For the full configuration reference and every skill option, see the Pasture Protocol repository on GitHub.