A lightweight, self-hosted AI agent designed for extreme efficiency, Picobot delivers advanced capabilities like persistent memory, tool execution, and chat integration within a single, ~8MB Go binary. It targets users seeking powerful AI agents on resource-constrained hardware, such as low-cost VPS, Raspberry Pis, or mobile devices, bypassing the bloat of traditional Python/Node.js frameworks.

How It Works

The agent operates via a straightforward message-passing architecture: an Agent Loop processes inbound messages, builds context from memory and skills, interacts with LLMs through OpenAI-compatible APIs, and executes a suite of tools (filesystem, exec, web, etc.). Replies are then routed back. This design prioritizes minimal dependencies, enabling millisecond cold starts and low RAM footprint, making it ideal for edge deployments.

Quick Start & Requirements

• Primary install / run command: Deploy via Docker (docker run or docker-compose) or build the single binary from source using Go (go build -o picobot ./cmd/picobot).
• Non-default prerequisites and dependencies: Requires an OpenAI-compatible API key (e.g., from OpenAI, OpenRouter, Ollama) and optionally a Telegram Bot Token for chat integration.
• Estimated setup time or resource footprint: Minimal; ~8MB binary size, ~28MB Docker image, ~10MB idle RAM, instant cold starts.
• Links: Detailed guides available in HOW_TO_START.md and CONFIG.md.

Highlighted Details

• Ultra-lightweight: Single binary (~8MB), zero runtime dependencies, runs on minimal hardware (256MB RAM).
• Extensive Toolset: Features 11 built-in tools for filesystem access, shell command execution, web fetching, messaging, subagent spawning, and task scheduling.
• Persistent Memory & Skills: Automatically organizes daily notes and long-term memory, supports semantic search, and allows creating/managing reusable skills as markdown files.
• Telegram Integration: Quick setup for conversational interaction via Telegram.

Maintenance & Community

The project roadmap outlines planned features like WhatsApp and Discord integration, alongside additional tools. Contributions are welcomed via issues and pull requests. Specific community channels or core maintainer details are not provided.

Licensing & Compatibility

MIT License, permitting unrestricted use, modification, and distribution, including for commercial purposes.

Limitations & Caveats

Current chat integration is limited to Telegram; support for other platforms like WhatsApp and Discord is planned. The toolset is extensive but may require further expansion for specific use cases not yet covered.