v1.3.0 — stable

ARGENTOR

The Secure AI Agent Framework — build production autonomous agents in Rust with sandboxed skills, 6-layer guardrails, and 8 LLM backends out of the box.

17
crates published
220K
lines of Rust
5,359
tests passing
8
LLM backends

Everything you need. Nothing you don't.

Each capability ships as a focused crate — include only what your agent needs.

🛡️

Security-First Guardrails

6-layer pipeline: PII detection, prompt injection, toxicity, shell injection, base64 exfiltration, and unicode smuggling — all before any LLM call.

🧠

Multimodal & Vision

Real HTTP multimodal calls to Claude, OpenAI, and Gemini. Pass images, audio, and documents. Streaming responses out of the box.

🌐

Web & Computer Use

Built-in web fetch, search, and extraction. Computer use skill for screenshot, mouse, and keyboard control. No external dependencies.

🤖

Multi-Agent Orchestration

Orchestrator-Workers pattern per Anthropic guidance. TaskQueue, AgentMonitor, agent profiles, and human-in-the-loop approval flows.

🐍

Python SDK

First-class Python bindings via PyO3. Run Argentor agents from Python scripts, notebooks, or FastAPI services — full feature parity.

🔌

MCP Native + WASM Skills

MCP client (JSON-RPC 2.0 stdio) ships as a core skill. WASM sandboxed plugins via Wasmtime — untrusted code runs isolated, always.

Quick Start

Run the hello-world example in under a minute — no API keys required for the demo backend.

# Clone and run the demo (uses DemoBackend, no API keys needed) git clone https://github.com/fboiero/Agentor && cd Agentor cargo run --example hello_world # Or add to Cargo.toml and point at a real LLM argentor-agent = "1.3" argentor-security = "1.3" # Run against Claude ANTHROPIC_API_KEY=sk-... cargo run --example hello_world -- --backend claude

Benchmark Highlights

Measured against a Python/LangChain baseline on equivalent pipelines. Full methodology →

11×
faster end-to-end latency
7.9×
lower cost per 1K tool calls
58.3%
guardrail block rate on red-team suite

Results from benchmarks/ — run cargo bench to reproduce on your hardware.