AgentRun is an open-source platform for running managed agents on top of a coding-agent harness such as Claude Code, Codex, OpenCode or pi.
Follow AgentRun to get an email when its listing changes, such as new pricing or features. One digest a day at most, and you can unsubscribe anytime.
It adds self-improvement, durable execution and a control plane: agents keep notes, skills and learned workflows from each run, runs survive restarts, and one console shows approvals, failures, spend and results.
- 01
Runs agents on Claude Code, Codex, OpenCode or pi harnesses with your own model keys
- 02
Learning mode that turns each run's notes, skills and workflows into graded commits in the agent's own git folder
- 03
Durable runs that carry on after the server is killed and restarted
- 04
Per-agent Docker sandbox, Linux desktop, Android phone or browser
- 05
Approvals, spend caps and a Mission Control console for every agent
- 06
Every agent exposed as an API, with a Playground to start runs
- 07
Templates, including a code reviewer you can tune on your own pull requests
Open Source
AgentRun is free and open-source under the Apache-2.0 licence and runs on your laptop, in one container or on your own server. You bring your own model keys (OpenRouter, Anthropic or OpenAI), so model usage is billed by those providers.
Best for
Teams that run the same agent jobs repeatedly, such as code review or research tasks, and want them to get cheaper and more reliable over time on their own infrastructure.
Not ideal for
Developers looking for an interactive coding assistant in the editor, or Windows users, since the quick start targets macOS and Linux.
A promising open-source control plane for repeat agent work, with a distinctive learning loop; still in beta, so expect fast changes.
Shown on Hacker News on November 12, 2026; v0.1.0-beta.36 was released on November 13, 2026, after beta.35 on November 12 and beta.34 on November 11.
As of
Is AgentRun free?+
Yes - AgentRun is Open Source. AgentRun is free and open-source under the Apache-2.0 licence and runs on your laptop, in one container or on your own server. You bring your own model keys (OpenRouter, Anthropic or OpenAI), so model usage is billed by those providers.
Is AgentRun open source?+
Yes - AgentRun is open source. AgentRun is free and open-source under the Apache-2.0 licence and runs on your laptop, in one container or on your own server. You bring your own model keys (OpenRouter, Anthropic or OpenAI), so model usage is billed by those providers.
Who is AgentRun best for?+
Teams that run the same agent jobs repeatedly, such as code review or research tasks, and want them to get cheaper and more reliable over time on their own infrastructure.
Who is AgentRun not ideal for?+
Developers looking for an interactive coding assistant in the editor, or Windows users, since the quick start targets macOS and Linux.
What are the best AgentRun alternatives?+
The closest AgentRun alternatives on ai.dosa.dev are Jules by Google, OpenHands, SWE-agent - all listed under Autonomous & Async Agents.
Who makes AgentRun?+
AgentRun is developed by Parcha. It is listed in the Autonomous & Async Agents category on ai.dosa.dev.
Favorite this tool to revisit it later, or Zap it to contribute to the public vote count.