Back to directory

clawd

A personal AI assistant that runs as a single always-on Swift daemon on your own machine, pairing a private Telegram bot with a chosen LLM to remember facts, run scheduled and proactive tasks, and execute approved tools.

clawd

clawd is a personal AI assistant packaged as one pure-Swift daemon that runs on hardware you own and talks to you through a private Telegram bot. It stores its state -- a SQLite database, encrypted secrets, and Markdown files -- in one directory you can edit by hand, and durable facts persist and are recalled by importance and recency.

Scheduled and proactive tasks fire on a cron-like schedule across restarts and time changes, and an opt-in heartbeat respects quiet hours. Consequential tool calls, such as writes or code execution, wait for explicit approval in Telegram, and a policy engine treats inbound content as untrusted data. Users can write reusable skills as SKILL.md files, and connect additional MCP servers whose tools join the built-ins at the lowest trust tier.

It works with any OpenAI-compatible model endpoint, including a ChatGPT subscription login, and can optionally delegate approved coding tasks to a local Codex installation, returning a result card and opening GitHub pull requests with the user's own repository access.

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.