This project provides a cognitive operating system designed to enable AI agents to autonomously interact with local system capabilities. It targets builders, researchers, and indie hackers seeking to empower AI agents with real-world task execution, abstracting complex system interactions into a unified protocol.

How It Works

llmbasedos exposes system capabilities (files, mail, APIs) via a Model Context Protocol (MCP), a JSON-RPC layer over UNIX sockets and WebSockets. This MCP acts as a unified abstraction, allowing LLM-agnostic interaction with various models (OpenAI, Gemini, LLaMA.cpp) and local services like file system access, email handling, and rclone synchronization. Agent workflows are implemented as Python scripts, leveraging an mcp_call() function for seamless integration, promoting flexibility and debuggability over traditional YAML-based approaches.

Quick Start & Requirements

• Install/Run: Clone the repo, organize llmbasedos_src/, add .env, lic.key, mail_accounts.yaml, and user files, then run docker compose build and docker compose up.
• Prerequisites: Docker and Docker Compose.
• Setup: Requires configuration of environment variables, license keys, and mail accounts.
• Links: Official Docs (implied by repo structure).

Highlighted Details

• Script-first Agent Logic: Agent workflows are Python scripts, not rigid YAML.
• LLM-Agnostic: Supports multiple LLM backends interchangeably.
• Local-first Security: Data remains local by default with virtual path jails.
• Unified Abstraction: All capabilities exposed as discoverable MCP methods.

Maintenance & Community

The project welcomes stars, forks, PRs, and experiments. Community engagement channels are not explicitly listed.

Licensing & Compatibility

The README does not explicitly state a license. The mention of "license tiers" and lic.key suggests a potential commercial or tiered licensing model, which may have restrictions on commercial use or closed-source linking.

Limitations & Caveats

The project is described as a "cognitive operating system" with a roadmap towards intention-based execution, implying it is still under active development and may not be production-ready for all use cases. The legacy YAML workflow engine is slated for deprecation.