LibreFang Overview

Relevant source files

• .github/workflows/update-contributors.yml
• .secrets.baseline
• CHANGELOG.md
• CONTRIBUTING.md
• Cargo.lock
• Cargo.toml
• README.md
• crates/librefang-desktop/tauri.conf.json
• docs/architecture/sidecar-channels.md
• i18n/README.de.md
• i18n/README.es.md
• i18n/README.fr.md
• i18n/README.ja.md
• i18n/README.ko.md
• i18n/README.md
• i18n/README.pl.md
• i18n/README.zh.md
• openapi.json
• packages/whatsapp-gateway/package.json
• sdk/javascript/package.json
• sdk/python/setup.py
• sdk/rust/Cargo.toml
• web/public/404.html
• web/public/assets/contributors.svg
• web/public/assets/star-history.svg
• xtask/baselines/openapi.sha256

LibreFang is an open-source Agent Operating System built from the ground up in Rust README.md35-37 Unlike traditional chatbot frameworks that wait for user input, LibreFang is designed to run autonomous agents that work 24/7 on schedules, monitoring targets, and performing complex tasks across multiple communication channels README.md39-40

The system is architected as a highly modular workspace consisting of 34+ members in the Rust workspace, providing a robust foundation for orchestration, long-term memory, and secure tool execution Cargo.toml3-34

System Capabilities

LibreFang provides a comprehensive suite of features for agent deployment and management:

• Autonomous "Hands": Pre-built capability packages (e.g., Researcher, Trader, DevOps) that run independently without constant prompting, defined by HAND.toml manifests README.md93-104
• Multi-Channel Integration: Over 40 messaging adapters supported via a sidecar-aware architecture README.md145-146 including platforms like Telegram, Discord, Slack, and WhatsApp CHANGELOG.md18-19
• LLM Agnostic: Supports a vast array of providers (Anthropic, Gemini, OpenAI, Ollama, etc.) via the librefang-llm-driver abstraction README.md129-130 Recent updates include support for DeepSeek v4-pro reasoning models CHANGELOG.md17-73
• Security First: Implements deep security layers, including WASM sandboxing via wasmtime, Merkle audit chains in librefang-runtime-audit, and taint tracking README.md118-182 Cargo.toml9-183
• Persistent Memory: Uses SQLite with WAL mode for session storage, vector embeddings for semantic search, and automated context compaction README.md117 Cargo.toml108 MCP server configurations are also persisted in SQLite to survive restarts CHANGELOG.md25-27
• Cross-Platform Interfaces: Includes a CLI, a React-based dashboard, and a Tauri 2.0 native desktop application README.md123-124 crates/librefang-desktop/tauri.conf.json3-5

High-Level Architecture

The following diagram illustrates how natural language interactions from various channels are processed by the core Rust entities and translated into autonomous actions.

From Natural Language to Code Execution

Sources: crates/librefang-kernel/src/kernel/mod.rs13-45 Cargo.toml3-34 CHANGELOG.md62-66

Workspace & Crate Structure

LibreFang is organized into a modular Rust workspace to separate concerns between the core engine, platform integrations, and user interfaces.

Crate Group	Purpose	Key Crates
Core	The heart of the OS; handles orchestration and execution logic.	librefang-kernel, librefang-runtime, librefang-types
Persistence	Manages long-term memory, vector storage, and migrations.	librefang-memory, librefang-memory-wiki, librefang-import
Interfaces	Provides ways for users and external systems to interact.	librefang-api, librefang-cli, librefang-desktop, librefang-http
Capabilities	Extends agent functionality through skills and autonomous hands.	librefang-skills, librefang-hands, librefang-extensions
Communication	Bridges the kernel to external messaging platforms and P2P.	librefang-channels, librefang-wire
Runtime Ext.	Specialized execution environments and protocols.	librefang-runtime-mcp, librefang-runtime-sandbox-docker, librefang-acp, librefang-rl-export

For a detailed map of the workspace and their dependencies, see Workspace & Crate Structure.

Sources: Cargo.toml3-34 README.md113-138

Subsystem Relationships

The system operates through a central LibreFangKernel which coordinates specialized subsystems to manage the lifecycle of agents and their environment crates/librefang-kernel/src/kernel/mod.rs1-12

Kernel Orchestration Flow

Sources: crates/librefang-kernel/src/kernel/mod.rs1-115 Cargo.toml3-34 CHANGELOG.md109

Getting Started

LibreFang can be installed via a shell script, Homebrew, Docker, or Cargo. Once installed, the librefang init command provides an interactive TUI wizard to configure LLM providers and basic settings README.md49-61

• Installation: curl -fsSL https://librefang.ai/install.sh | sh README.md51
• Initialization: librefang init README.md60
• Starting the OS: librefang start (starts the daemon and opens the dashboard at http://localhost:4545) README.md57

For detailed installation guides and first-run configuration, see Getting Started.

Sources: README.md49-85

Child Pages

• Getting Started: Installation methods (Shell, Homebrew, Docker, npm/pip/Cargo), the init wizard TUI, and daemon lifecycle management.
• Workspace & Crate Structure: A deep dive into the Rust workspace, crate responsibilities, and the internal dependency graph including librefang-types, librefang-kernel, and librefang-runtime.