The Orchestration Layer for Autonomous Intelligence.
한국어 | 简体中文 | 日本語 | Français | Español | Deutsch | Português
LibrAgent is a high-performance Meta Agent Platform designed to industrialize autonomous workflows. Moving beyond simple chat interfaces, it provides a robust orchestration engine and a secure execution substrate where specialized agents collaborate to solve complex, multi-step missions.
By implementing open standards like the Model Context Protocol (MCP) and a recursive delegation architecture, LibrAgent transforms raw LLM capabilities into a coordinated swarm of intelligence.
Modern AI work requires more than a stateless window; it requires Strategic Autonomy. LibrAgent bridges the gap between manual prompts and fully autonomous systems by providing a local-first environment where humans can design, deploy, and govern agentic teams with precision.
Recursive delegation and high-fidelity tool usage in a unified, stateful substrate.
LibrAgent is built for scale. It allows agents to spawn, brief, and manage specialized sub-agents with strict governance.
- Hierarchical Delegation: Transparent parent-child lineages with configurable depth and fan-out limits.
- Role-Based Specialization: Define unique "Souls" and toolsets for specific mission phases.
- Swarm Coordination: Real-time message routing and terminal results monitoring across the agent tree.
Standardization is at the heart of the platform. We use the Model Context Protocol to ensure infinite extensibility.
- Universal Tooling: Instantly connect to any MCP-compliant server (GitHub, Brave, Slack, etc.).
- Dynamic Service Proxying: Isolated tool instances per session to prevent context leakage.
- One-Click Integration: A curated catalog of essential agent capabilities.
Agents operate within a long-lived environment that preserves the state of their work.
- Shared Workspace: A secure, persistent file substrate where all agents in a lineage can collaborate.
- Live Execution Environment: Persistent browser sessions (Tauri) and sandboxed shells (Python/Node.js) that stay alive between turns.
- Context Compaction: Intelligent history management for sustained performance in long-running missions.
Take control of autonomous execution with robust safety and scheduling features.
- YOLO Mode: Configurable autonomous execution for high-trust environments.
- Scheduled Missions: CRON-based automation with automatic recovery and workspace targeting.
- Observability: Real-time performance metrics (TPS) and prompt caching transparency.
LibrAgent is an industrial-grade platform. Explore our detailed resources:
- Navigation Guide: Explore the Command & Control hub, including
/assistants(Role Definitions) and/playbooks(Workflow Blueprints). - Architecture Guide: Deep dive into the orchestration engine and session isolation logic.
Download the latest binaries for Windows, macOS, or Linux from the Release page.
Developer Setup:
git clone https://github.com/fritzprix/libr-agent
cd libr-agent
pnpm install
pnpm tauri dev- Local First: Your data, keys, and agent "souls" remain under your exclusive control.
- Memory Safety: Powered by Rust and Tauri for maximum security and performance.
- Open Standards: Fully committed to MCP and open-source interoperability.
We are building the future of autonomous intelligence. Join us on GitHub.
License: MIT