Back to directory

exoclaw

A minimal, protocol-only Python framework that adds tool use, conversation memory and any LLM provider to an existing application, kept to roughly 2,000 auditable lines with one runtime dependency.

exoclaw

exoclaw is a fork of the nanobot project, cut down to just the agent loop and the protocols around it, so that storage, channels, tools and providers each live in separate packages a developer opts into rather than a bundle of built-in integrations. Its authors built it because every baked-in feature in the original became a release blocker, where an unrelated integration's API change could hold up an unrelated bug fix.

It installs with pip, depends only on the structlog library at runtime, and processes messages through a loop that moves an inbound message through a bus to the agent loop, on to the configured LLM provider and tools, and back out through a channel. A companion package, exoclaw-nanobot, bundles a default provider, conversation, tools and an interactive CLI channel into a working bot, and a separate plugin catalogue adds integrations such as Slack, Telegram, Discord and Matrix. A Go port is forward-ported periodically from tagged Python releases.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.