Hmbown/CodeWhale · error

repeated coordination interrupt

Error message

repeated coordination interrupt

What it means

This is a test panic from `Option::expect`/`Result::expect`: `interrupt_child` returned None/Err on the second call, so the test asserting that repeated coordination interrupts from the same parent succeed never got a value. The library throws expect panics like this when an internal manager API that the test contract requires to succeed does not.

Solutions

1. Ensure `interrupt_child` is idempotent: a repeated interrupt on an already-Interrupted agent returns the same checkpoint rather than failing.
2. Check that the agent_id was inserted into `manager.agents` before the first interrupt and was not removed by the first interrupt.
3. Run the failing test to inspect which call returned None/Err and diff against recent changes in the interrupt implementation.

Example fix

// before: second call errors on already-interrupted agent
// after: idempotent interrupt in manager
if matches!(self.agents.get(id).map(|a| &a.status), Some(SubAgentStatus::Interrupted(_))) {
    return self.last_interrupt_checkpoint(id).ok_or_else(|| anyhow!("unknown agent"));
}

// before calling
assert!(manager.agents.contains_key(&agent_id), "agent must exist before interrupt");

fn is_interrupted(s: &SubAgentStatus) -> bool { matches!(s, SubAgentStatus::Interrupted(_)) }

match manager.interrupt_child(&agent_id, Some("agent_parent"), reason.clone()) {
    Ok(cp) => cp,
    Err(e) => panic!("repeated coordination interrupt failed: {e:#}"),
}

When it happens

Trigger: Calling `SubAgentManager::interrupt_child(&agent_id, Some("agent_parent"), reason)` a second time for an already-interrupted or non-existent agent, when the implementation fails to treat a repeated interrupt as idempotent.

Common situations: Regression in the subagent manager's interrupt path after refactoring terminal-state transitions; running the subagent test suite in crates/tui/src/tools/subagent/tests.rs after changing status handling so the second interrupt errors instead of returning the same Interrupted checkpoint.

Understand the failure class

Background: "This is a bug, please report it": internal invariant violations, unreachable panics, and SNH errors explained — this error's family across 47 libraries.

Related errors

• parent interruption fan-in
• cancel wins race
• cancelled snapshot exists
• Expected Ok for scoped npm package via
• Expected Ok when cwd anchors relative path, got

AI-assisted analysis of Hmbown/CodeWhale@73e0f67d83 (2026-09-22). Data as JSON: /api/errors/75035539cda40d23. Report an issue: GitHub.

Appendix: source

Thrown at crates/tui/src/tools/subagent/tests.rs:6708

                .interrupt_child(&agent_id, Some(caller), "forbidden".into())
                .is_err()
        );
    }
    manager.record_worker_event(
        &agent_id,
        AgentWorkerStatus::RunningTool,
        Some("step 2/8: running tool 'read_file'".to_string()),
        Some(2),
        Some("read_file".to_string()),
    );

    let reason = "parent rerouted this lane".to_string();
    let (prior, first) = manager
        .interrupt_child(&agent_id, Some("agent_parent"), reason.clone())
        .expect("first coordination interrupt");
    let (_, second) = manager
        .interrupt_child(&agent_id, Some("agent_parent"), reason.clone())
        .expect("repeated coordination interrupt");
    assert_eq!(prior.status, SubAgentStatus::Running);
    assert!(matches!(
        first.status,
        SubAgentStatus::Interrupted(ref actual) if actual == &reason
    ));
    assert_eq!(second.status, first.status);
    assert_eq!(
        first
            .checkpoint
            .as_ref()
            .map(|checkpoint| (checkpoint.reason.as_str(), checkpoint.steps_taken)),
        Some(("test_checkpoint", 2))
    );

    let completion = completion_rx
        .try_recv()
        .expect("parent interruption fan-in");
    assert!(completion.payload.contains(r#""status":"interrupted""#));

View on GitHub (pinned to 73e0f67d83)