Agent Autoloop
Agent Autoloop is an open-source tool that runs command-line AI coding agents in the background on queued tasks and asks you in chat only when unsure.
No visits from OpenBeat in the last 30 days yet. Every click on “Visit” is counted here.
Pricing and API checked by an OpenBeat editor on Nov 11, 2026 · facts from the maker’s own site ·
Tool insights
Agent Autoloop lets command-line agents such as Claude Code or Codex work through queued tasks without a terminal staying open. When an agent needs a decision, it sends a card to a chat app and waits, using no process or tokens until you answer.
CLI agents normally need someone watching a terminal, and they stall or guess when they hit a question. Agent Autoloop moves that work into the background and turns questions into cards you can answer later.
Good for developers and small teams who already use a CLI coding agent and want it to handle recurring jobs such as reviews, merges and inbox triage. Less suited to anyone who does not want to install Node and Git, or who needs a hosted service with a vendor behind it.
What it does
Agent Autoloop is an open-source runner for command-line AI agents. It takes events, queues them as tasks and hands them to an agent installed on your machine. The agent runs in the background, so no terminal window has to stay open.
When the agent is unsure, Agent Autoloop sends a decision card to a chat channel. The run is suspended while it waits. After you answer, the original agent session resumes with your choice.
Key features
- Runs CLI agents in the background without an open terminal.
- Dispatches events from cron, GitHub and webhooks to agents.
- Sends decision cards to Feishu, DingTalk, WeCom, Slack and Discord.
- Suspends waiting runs without using a process or tokens.
- Resumes the original agent session after an answer.
- Keeps an event log, with a command that explains each run.
- Offers a delivery pipeline: review, wait for CI, merge, then check the deploy.
- Lets you pause, resume and set run rules from the command line.
- Has a plugin system for events, agents, channels and delivery.
Listed agent integrations are Claude Code, Codex, Gemini CLI, OpenCode and Cursor. GitHub and GitLab are also listed.
Pricing
As of November 2026, Agent Autoloop is open source under the MIT license. No paid plans or prices are stated on the maker's site. Whether a hosted or free plan exists is not stated.
API and integrations
The maker lists an API, but the facts given do not describe it in detail. No SDKs are stated. Integrations include the chat apps Feishu, DingTalk, WeCom, Slack and Discord, the code hosts GitHub and GitLab, and the agents named above. Platforms are Windows, macOS and Linux, used through the command line. Documentation and releases are on GitHub.
Good to know
- It needs Node 24 or newer and Git.
- It needs a command-line agent installed locally and already logged in.
- Text channels need hub.publicUrl set to an address that can be reached.
- The maker has not tested durability after a power loss.
- Multi-turn ACP sessions for Gemini CLI and OpenCode have not yet been run.
- The maker has not yet run several of the listed agent commands.
- Recent releases (v1.6 to v1.8.1) mostly add fixes and review and task-limit settings. Release dates are not stated.
FAQ
Is Agent Autoloop free?
As of November 2026, Agent Autoloop is open source under the MIT license. The maker's site states no paid plans or prices. Whether a hosted free plan exists is not stated.
Which AI agents does Agent Autoloop work with?
It lists Claude Code, Codex, Gemini CLI, OpenCode and Cursor. The maker has not yet run several of the listed agent commands. Multi-turn ACP sessions for Gemini CLI and OpenCode are also untested.
What do I need to run Agent Autoloop?
You need Node 24 or newer, Git and a logged-in command-line agent installed locally. It runs on Windows, macOS and Linux through the command line. Text chat channels also need hub.publicUrl set to a reachable address.
How does Agent Autoloop ask for decisions?
It sends a decision card to Feishu, DingTalk, WeCom, Slack or Discord when an agent is unsure. The run is suspended and uses no process or tokens while it waits. After you answer, the original agent session resumes.
Can Agent Autoloop merge code automatically?
It has a delivery pipeline that covers review, waiting for CI, merging and checking the deploy. It works with GitHub and GitLab. How much of this is automatic depends on your rules, which the facts do not detail.
Releases
No tracked releases yet.
In the news
No coverage yet.