Free
Stop risky calls, keep the job running, and keep the record.
- Dangerous calls stopped before they run: cloud teardowns and other things you can't undo, with a model checking the calls our rules can't decide.
- A way forward: a stopped call comes back with a safer way to do it, so the job can still finish.
- A person on the risky calls: a payment can wait until someone on your team approves it on the website.
- Runaway runs cut off: an agent stuck repeating the same call gets stopped.
- One set of team rules: set a rule once, and Claude Code asks first on every teammate's machine.
- Evidence for your security review: download a file with the rules in force and the calls that were stopped or held.
For TypeScript tools, the gateway is what stops a call.
Use our hosted gateway with a free key, or run it yourself with Docker.
Use our hosted gateway.
Nothing to run. Add a free key to your project, and our gateway checks your Python or TypeScript tools before they run. The key includes up to 100 model checks a day, for calls our rules can't decide.
Or run it yourself, with Docker.
Using the TypeScript SDK? To block TypeScript tools, wrap them with agentxProtect and set AGENTX_API_KEY (and AGENTX_GATEWAY_URL for a gateway that is not on this machine): the gateway checks each call before it runs. If it cannot answer, the call runs unchecked unless AGENTX_FAIL_MODE=closed.
Stuck? Email founders@agentx-core.com and we'll help you get it running.
Want one set of rules for your whole team's coding agents, and to approve risky calls from one page? The same free key runs that too. See how