What causes "Codewhale messages request should succeed" in Hmbown/CodeWhale?

Check the mock server's registered routes match the request path/method the client actually sends

Codewhale messages request should succeed

Error message

Codewhale messages request should succeed

What it means

This `expect` panics when `client.create_message(...)` returns `Err` while sending an Anthropic-messages-format request through the Codewhale client, aborting the test with 'Codewhale messages request should succeed'. The client retries/fails the HTTP call and the test treats any error as a hard failure because the mock server is configured to always succeed.

Solutions

Check the mock server's registered routes match the request path/method the client actually sends
Assert on the returned `Err` payload to see the HTTP status or transport error
Verify bearer token and base URL setup in the `codewhale_client` helper
Ensure required headers (e.g. anthropic-version) are accepted by the mock

Example fix

// before
client.create_message(minimal_zen_request("anthropic/claude-sonnet-5")).await
    .expect("Codewhale messages request should succeed");
// after
let resp = client.create_message(minimal_zen_request("anthropic/claude-sonnet-5")).await
    .unwrap_or_else(|e| panic!("Codewhale messages request should succeed: {e:?}"));

Defensive patterns

Strategy: try-catch

Validate before calling

assert_eq!(client.wire_format, WireFormat::AnthropicMessages);
// ensure a mock route exists for the exact method+path before calling

Type guard

fn expect_ok<T>(r: Result<T, impl std::fmt::Debug>) -> T {
    r.unwrap_or_else(|e| panic!("messages request failed: {e:?}"))
}

Try / catch

match client.create_message(req).await {
    Ok(resp) => resp,
    Err(e) => panic!("Codewhale messages request should succeed: {e:?}"),
}

Prevention

Register mock routes before issuing client requests
Assert wire_format before calling to catch routing mistakes
Log client errors instead of bare expect for debuggability

When it happens

Trigger: Calling `create_message` on a `CodewhaleClient` with `wire_format == WireFormat::AnthropicMessages` (crates/tui/src/client.rs:7841) when the mock server returns a non-2xx status, the request cannot be built, the connection fails, or all retries are exhausted.

Common situations: Mock route not registered so wiremock returns 404; missing `anthropic-version` handling; wrong base URL or API key in the test config; TLS or socket errors in CI.

Understand the failure class

Background: 'Something went wrong' / 'Request failed (500)' / 'HTTP error! status: 404' — what failed HTTP requests actually mean and how to find the real cause — this error's family across 28 libraries.

Related errors

AI-assisted analysis of Hmbown/CodeWhale@73e0f67d83 (2026-09-22). Data as JSON: /api/errors/34050dee65c5d27c. Report an issue: GitHub.

Appendix: source

Thrown at crates/tui/src/client.rs:7841

                "id": "msg_cw",
                "type": "message",
                "role": "assistant",
                "content": [{"type": "text", "text": "ok"}],
                "model": "anthropic/claude-sonnet-5",
                "stop_reason": "end_turn",
                "stop_sequence": null,
                "usage": {"input_tokens": 1, "output_tokens": 1}
            })))
            .expect(1)
            .mount(&server)
            .await;

        let client = codewhale_client(&server, "anthropic/claude-sonnet-5");
        assert_eq!(client.wire_format, WireFormat::AnthropicMessages);
        client
            .create_message(minimal_zen_request("anthropic/claude-sonnet-5"))
            .await
            .expect("Codewhale messages request should succeed");

        let requests = server.received_requests().await.expect("recorded request");
        assert_eq!(requests.len(), 1);
        assert_codewhale_bearer(&requests[0]);
        assert_eq!(
            requests[0]
                .headers
                .get("anthropic-version")
                .and_then(|value| value.to_str().ok()),
            Some("2023-06-01")
        );
    }

    /// A catalog row stating `codewhale.protocol = "responses"` must dispatch
    /// to the account API's Responses surface — `{base}/responses` — not the
    /// Chat Completions default its `openai/` namespace alone would imply.
    #[tokio::test]
    async fn codewhale_responses_catalog_row_dispatches_to_responses_endpoint() {

View on GitHub (pinned to 73e0f67d83)