Skip to main content

run_live_scenario

forge_test_harness::sdk_wire

Function run_live_scenario 

Source 
pub async fn run_live_scenario<F, Fut>(
    scenario: &str,
    options: Options,
    drive: F,
) -> Result<Option<()>, Error>
where
    F: FnOnce(Client, ClientEvents) -> Fut,
    Fut: Future<Output = Result<(Client, ClientEvents), Error>>,
Expand description

Run a live-capture scenario end-to-end: build options, spawn a recorded claude, drive the scenario to a Result frame, dump the trace to target/wire-traces/, assert every inbound line decodes cleanly.

Caller supplies:

  • scenario: a short slug (e.g. "bash_tool") used in trace filenames.
  • options: fully-built forge_sdk::Options - set tools, permission_mode, hooks, MCP servers, etc. here.
  • drive: async closure that drives the scenario once the client is ready. Typically calls send_user_message(...) and may register turn-specific state.

§Skip semantics

When FORGE_WIRE_CAPTURE is unset, returns Ok(None) immediately without touching the network - scenarios compile and link in CI but only run when the developer opts in.

§Errors

Any forge_sdk::Error surfaced during spawn / drive / drain. Panics on trace-dump failure or decode-completeness failure - those are harness invariants, not recoverable errors.

§Panics

  • Transport panics are propagated.
  • Trace file I/O failure.
  • Decode-completeness regression in the captured inbound frames.