Back to directory

NanoClaw

A lightweight, container-isolated alternative to OpenClaw: an AI assistant that runs each agent in its own Linux container and connects to Slack, Telegram, Discord, WhatsApp and other messaging apps.

NanoClaw

Where OpenClaw runs everything in one Node process protected mainly by application-level allowlists and pairing codes, NanoClaw runs each agent in its own Linux container with filesystem isolation, so an agent's shell access is confined to that container rather than the host, and the whole codebase is kept small enough -- one process and a handful of files -- to read directly. Setup provisions each Slack agent its own Slack app, avatar and workspace install without manually creating tokens, and new teammates spawned from chat get their own bot identity, container and memory while sharing rooms and canvases.

A setup script installs Node, pnpm and Docker if missing, installs a credential gateway, registers an Anthropic credential, builds the agent container, and pairs a first channel such as Slack, Telegram, Discord, WhatsApp, iMessage or a local CLI; a separate migration script moves an existing v1 install's state, groups, sessions and channel authentication into a new v2 checkout.

It is written in TypeScript, runs on Anthropic's Agents SDK, and is licensed under the MIT license.

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.