Repository navigation
Cowork-to-Code Bridge — give SuperAGI agents an execution backend on the user's real machine #1578
abhinaykrupa
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Hi all — sharing a small open-source tool that pairs well with SuperAGI when an agent needs the user's actual computer rather than a sandbox.
Cowork-to-Code Bridge (MIT, Python, zero deps) is a local-first async RPC layer: an agent queues a task as a JSON file, a daemon on the user's Mac/Linux box executes it (optionally handing free-form tasks to a local Claude Code agent), and the result comes back through a shared directory. No open ports, no cloud relay.
Why it might interest SuperAGI users:
queue_task/poll_task_resultwith idempotency keys, so retries never double-fire a deploy.max_budget_usd, permission scopes (readonly → full), and model-tier routing, with owner-side hard ceilings.Integration notes for external frameworks are in docs/EXTERNAL_AGENT_INTEGRATION.md. Feedback very welcome — especially on what a first-class SuperAGI toolkit wrapper should look like.
All reactions