sandbox-agent/server/packages/sandbox-agent/tests/sessions/reasoning.rs

50 lines
1.7 KiB
Rust

// Reasoning capability checks are isolated from baseline snapshots.
include!("../common/http.rs");
fn reasoning_prompt(_agent: AgentId) -> &'static str {
"Answer briefly and include your reasoning."
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn reasoning_events_present() {
let configs = test_agents_from_env().expect("configure SANDBOX_TEST_AGENTS or install agents");
for config in &configs {
let app = TestApp::new();
let capabilities = fetch_capabilities(&app.app).await;
let caps = capabilities
.get(config.agent.as_str())
.expect("capabilities missing");
if !caps.reasoning {
continue;
}
let _guard = apply_credentials(&config.credentials);
install_agent(&app.app, config.agent).await;
let session_id = format!("reasoning-{}", config.agent.as_str());
create_session(&app.app, config.agent, &session_id, test_permission_mode(config.agent))
.await;
let status = send_status(
&app.app,
Method::POST,
&format!("/v1/sessions/{session_id}/messages"),
Some(json!({ "message": reasoning_prompt(config.agent) })),
)
.await;
assert_eq!(status, StatusCode::NO_CONTENT, "send reasoning prompt");
let events = poll_events_until_match(
&app.app,
&session_id,
Duration::from_secs(120),
|events| events_have_content_type(events, "reasoning") || events.iter().any(is_error_event),
)
.await;
assert!(
events_have_content_type(&events, "reasoning"),
"expected reasoning content for {}",
config.agent
);
}
}