AgenticSeek

AgenticSeek

★ 4.4 🔖 7000 saves 💰 Free ChatbotsGitHub

AgenticSeek is a fully local, voice-enabled autonomous AI agent: a Manus-style assistant that runs entirely on your own hardware instead of in the cloud. It can browse the web, write and run code, read files and plan multi-step tasks, while every model call happens on your machine through local runtimes such as Ollama or LM Studio. With roughly 27,500 stars on GitHub, it is one of the best-known open-source attempts at a private, self-hosted general agent.

Overview

Most general-purpose agents send your prompts, files and browsing activity to a third-party cloud. AgenticSeek takes the opposite approach: the agent loop, its tools and the language model all run locally, so nothing leaves your computer unless you deliberately connect an external API provider, which is also supported. You interact with it by voice or text, assign a goal, and the agent breaks the goal into steps, researches topics online, writes code to solve sub-problems and reports back. Because it is model-agnostic, you can pair it with capable local models for full privacy or with hosted models when a task needs more power, trading off privacy and capability per setup.

Key Features

  • 100% local operation with Ollama, LM Studio or other local model servers, with optional API providers
  • Voice interaction for hands-free control of the agent
  • Autonomous web browsing and research to gather information for a task
  • Code writing and execution as part of multi-step task plans
  • Task planning that decomposes goals and routes work across specialized sub-agents
  • Privacy by design: prompts and files stay on your machine in local mode

💳 Pricing

AgenticSeek is free and open source under the GPL-3.0 license. There are no subscriptions; your only costs are your own hardware and, if you choose them, any third-party model API keys.

Who Is It For

AgenticSeek is for privacy-conscious users, researchers and tinkerers who want a general-purpose autonomous agent they fully control, and for anyone evaluating how far local open-weight models can go on real agentic tasks.

🚀 Visit AgenticSeek Official Site

🚩 Report an issue