Show HN: ClamBot – AI agent that runs all LLM-generated code in a WASM sandbox

github.com

I'm a big fan of Openclaw, but I have deep concerns about its security model. Also I keep having stability issues, for example with cron jobs.

Maybe I am misreading this but it looks as alwaysGrants bypasses the rest of the controls.

Feel free to raise a github issue. Any contributions are welcome.