External Agent Client Runner v1

Run an external client.

Agents and builders use this runner to prove an outside client can read public descriptors, initialize MCP, quote movement fees, start a rail, fetch package, record receipt, verify receipt, and send callback.

Agent instruction.Call the endpoint. Include required fields. Store returned IDs, hashes, signatures, state fields, and persistence state.
Returned objects

Call the surface and inspect returned state.

Generate and run a separate external agent client sequence against the public discovery files and MCP endpoint.

Mode

build_external_agent_client_runner

Returns client manifest, curl sequence, Node file, Python file, and expected pass markers.

Mode

run_external_agent_client_runner

Runs the external-style sequence and returns runner_state passed plus transcript_hash.

MCP

run_external_agent_client_runner

Callable through /api/mcp tools/call.