AutoGPT is an open-source platform for building, running, and scheduling AI agents that carry out multi-step workflows from plain-English descriptions or a visual builder.
mindsdb/anton
Open-source self-improving agent harness that takes outcome-based tasks like reporting, email cleanup and integrations, running in a terminal with any LLM.
- Stars
- 757
- Forks
- 130
- Open issues
- 36
- Last commit
- yesterday
What is it?
Anton is an open-source agent harness that the MindsHub team uses as the default agent in its MindsHub Cowork workspace, and it can also run standalone from the terminal. You describe the outcome you want, such as a portfolio dashboard, a cleaned inbox, or a scheduled report, and Anton works out the steps itself.
Rather than relying on a fixed set of built-in connectors, Anton writes code on the fly in an isolated execution scratchpad, and it can build new integrations when none exists. It keeps a credential vault so secrets are not exposed to the LLM, and it stores session, semantic and long-term memory in a per-workspace .anton folder to improve on recurring workflows.
Anton includes web search and fetch tools. Their execution depends on the LLM provider: Anthropic and OpenAI run them natively, while other OpenAI-compatible endpoints can use Exa.ai or Brave as the search backend. The project is MIT-licensed and written in Python.
Key features
- 01Runs in a terminal with any LLM, or as the default agent in MindsHub Cowork.
- 02Writes and executes code in an isolated scratchpad to complete tasks it has no built-in skill for.
- 03Keeps secrets in a credential vault so they are not exposed to the model.
- 04Stores session, semantic and long-term memory in a per-workspace folder.
- 05Provides web search and fetch tools with provider-specific execution paths.
Use cases
- Generating a portfolio dashboard from live stock prices and performance data.
- Triaging a large email inbox to find unsubscribable marketing and cold outreach.
- Building a scheduled report that pulls from several databases.
- Prototyping a messaging integration so you can chat with an agent from a phone app.
Quick start
- 1
Install Anton on macOS or Linux
Install with the official instructions in the README - 2
Install Anton on Windows (PowerShell)
Install with the official instructions in the README - 3
Run Anton in your terminal
anton
Things to know
- Web search via generic OpenAI-compatible endpoints requires a separate Exa.ai or Brave key set up with anton setup-search.
- Analytics are enabled by default and must be turned off with ANTON_ANALYTICS_ENABLED=false if unwanted.
- The Windows scratchpad may need a manual firewall rule to reach the internet.
Topics
FAQ
- What is anton?
- Open-source self-improving agent harness that takes outcome-based tasks like reporting, email cleanup and integrations, running in a terminal with any LLM. Anton is an open-source agent harness that the MindsHub team uses as the default agent in its MindsHub Cowork workspace, and it can also run standalone from the terminal. You describe the outcome you want, such as a portfolio dashboard, a cleaned inbox, or a scheduled report, and Anton works out the steps itself.
- Is anton free to use? What license does it have?
- anton is open source under the MIT license. Check the license terms before using it in commercial products.
- Is anton still maintained?
- Actively maintained. The last commit to anton was on Oct 19, 2026.
- What can I use anton for?
- Generating a portfolio dashboard from live stock prices and performance data. Triaging a large email inbox to find unsubscribable marketing and cold outreach. Building a scheduled report that pulls from several databases. Prototyping a messaging integration so you can chat with an agent from a phone app.
- How do I install anton?
- Install Anton on macOS or Linux Install Anton on Windows (PowerShell) Run Anton in your terminal: anton
- Where can I get anton?
- The source code and installation instructions are on GitHub: https://github.com/mindsdb/anton
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source Python framework for orchestrating role-based autonomous AI agents (Crews) and event-driven workflows (Flows).
Python framework for building autonomous AI agents that research, plan, code and run tasks, with memory, RAG and 100+ LLM providers.
Fair-code workflow automation platform for building AI agents and workflows, combining a visual canvas with custom code and 1500+ integrations.
A Python framework and runtime for building, serving, and managing AI agent platforms, with a web UI for administration.
Open-source infrastructure for running AI-generated code in isolated cloud sandboxes, with JavaScript and Python SDKs.