lukacf/meerkat

lukacfTLA15updated 4mo ago

Meerkat - A modular, high-performance agent harness built in Rust.

ClaudeCodexCursorMCP

agenticaiai-agentsclaudeclaude-codeclicodexcodex-cligeminigemini-cliharnessmcp

README

<img src=".github/meerkat-logo.png" alt="Meerkat" width="280"> <h1 align="center">Meerkat</h1> A modular, high-performance agent harness built in Rust. <a href="#quick-start">Quick Start</a> • <a href="#capabilities">Capabilities</a> • <a href="#surfaces">Surfaces</a> • <a href="#examples">Examples</a> • <a href="https://docs.rkat.ai">Docs</a> <img src="https://img.shields.io/badge/Rust-1.94+-orange?logo=rust" alt="Rust 1.94+"> <img src="https://img.shields.io/badge/License-MIT%2FApache--2.0-blue" alt="License">

Why Meerkat?

Meerkat is a library-first, high-performance, modular agent harness -- composable Rust crates that handle the hard parts of building agentic systems: state machines, retries, budgets, streaming, tool execution, MCP integration, and multi-agent coordination.

That harness is backed by a shared runtime. The same sessions, tools, credentials, schedules, live channels, blobs, and mob members work across the CLI, services, SDKs, and browser/WASM delivery instead of each surface reimplementing agent behavior.

It is designed to be stable (typed session events, explicit terminal results, resumable persistence, scoped credentials) and fast (<10ms cold start, ~20MB memory, small standalone binaries for the common surfaces). Meerkat lifecycle flows are specified as typed formal state machines and mathematically proven with TLA+ where it matters, which means the system avoids getting stuck in invalid or unknown states.

The library still comes first; surfaces come second. Pick the entry point that fits your architecture: embed the crates directly, run a CLI task, host REST or JSON-RPC, expose MCP tools, script from Python or TypeScript, or ship a browser-delivered agent with @rkat/web.

How it compares

| | Meerkat | Claude Code / Codex CLI / Gemini CLI | |---|---|---| | **Desig