mirror of
https://github.com/fabro-sh/fabro.git
synced 2026-10-14 04:12:08 +00:00
The five twin-mode tests in `tests/it/workflow/hooks.rs` have not tested hooks since runs moved into the server. They wrote their `[[run.hooks]]` entries into the user's settings file, which `fabro run` no longer transmits, so no hook ran; and even with the hooks moved into a workflow config, the run executes in the isolated test server, which never learns the twin's base URL and calls the real OpenAI API with the twin namespace as its key. Prompt hooks fail open on an LLM error, so the tests that expect the run to proceed passed for the wrong reason, and the one that expects a block never could. Removing them rather than leaving ignored tests that pass or fail for unrelated reasons. Bringing hook coverage back needs the test harness to hand the isolated server the twin's base URL through its runtime environment; the pull request records what that involves. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| src | ||
| tests | ||
| build.rs | ||
| Cargo.toml | ||