AgentAPI

You will want this the day something that is not a keyboard needs to drive your agent. The trait that decides it in daily use is restraint: it relays messages and nothing else, so everything you already liked about the agent underneath survives the wrapper intact. Eleven agents are documented, from Amazon Q to Cursor CLI, which means the caller you write today does not get thrown away when you change your mind about which agent is best.

Pick it if the agent choice has to stay open. Pick the Claude Agent SDK when you have already committed to one vendor and want a supported client instead of a bridge.

The failure mode is screen scraping. Keystrokes go in, rendered terminal output is read back out as structure, and that works exactly as long as the rendering is stable. The agents being wrapped ship weekly and none of them owes this project a stable presentation layer. When the parse slips, what comes back is well-formed and wrong, which costs more to detect than an outright error.

What it does right is scope. File editing, git operations and isolation are all recorded as belonging to the wrapped agent, not to this. It relays and it does not pretend otherwise.

Three points. 1. The contract is published as a machine-readable schema in-tree with browsable docs beside the running process, so a client is generated rather than guessed. 2. Four endpoints, message list, message post, status, and an event stream, is a small surface, and small surfaces are exhaustively testable. 3. Nothing is claimed on any evaluation, which is correct for a transport.

The principled choice is refusing to own context assembly or edit application. Correctness here reduces to whether the emitted stream matches the declared schema, which a reader can verify in an afternoon rather than trust.

Nobody builds a company on a permissively licensed HTTP wrapper, and Coder is not trying to. The strategic logic is clean: their business is hosted development environments, and free permissive plumbing makes those environments the natural place agents run. Pricing power here is zero by design and that is the design working.

Switching cost accrues upstream in Coder's platform rather than in this repository, so the risk to a buyer is neglect rather than a pivot; strategic components stay maintained until the strategy moves. Likely acquirer of the parent is a cloud IDE vendor. Position: adopt freely, and track Coder's roadmap, not this one.

The demo is a curl command, which I appreciate. Sixty seats cost nothing and the meter belongs to whichever agent sits behind it. This is the unusual case that fits our pipelines natively, because an HTTP process is something my platform team already knows how to run behind its own authentication and its own logging. What the row will not give me is identity: no SSO, no SCIM, no audit log, because this is a component rather than a product.

That is acceptable inside our own perimeter and unacceptable outside it. Approved with conditions: internal network only, and one named team owns the deployment.

MIT and a single binary I curl straight out of releases, then agentapi server -- claude and the agent is on a port. That is the entire ceremony. My provider credentials never move; they stay with the agent underneath, so nothing new is holding keys. MCP is not implemented, and instead of hand-waving the README shows how to make this the backend of a server I write, which is honest and also more work.

A fork stays maintainable because the surface is tiny. I would rather have this than a vendor SDK, because I can put agents behind it that its authors never listed.