Summary
gateway event mute and gateway event cancel print a success message and exit 0 on an event where they cannot have had any effect, and reading the event back shows nothing changed.
Reproduce
On an event in a terminal FAILED state:
hookdeck gateway event mute <id> → Event <id> muted. exit 0
hookdeck gateway event cancel <id> → Event <id> cancelled. exit 0
hookdeck gateway event get <id> --output json
→ no field changed
For contrast, gateway event retry on the same event genuinely worked — attempts went 1 → 2.
Assessment
This is plausibly a legitimate server-side no-op: a terminal event cannot be cancelled or muted. The defect is that the CLI asserts it happened. Event <id> muted. is a statement about state that is not true.
Expected
Either report that the event was not in a state where the action applies, or report what the API actually did. A confirmation message should not be printed unconditionally on a 2xx that changed nothing.
Scope
Pre-existing, not a regression. Verified identical on fbb924a and the merged branch.
Why it matters
Same shape as several defects fixed in #392: the command reports success while nothing happened. Someone muting a noisy event believes it is muted.
Filed by Claude on Phil's behalf, from release-candidate regression testing.
Summary
gateway event muteandgateway event cancelprint a success message and exit 0 on an event where they cannot have had any effect, and reading the event back shows nothing changed.Reproduce
On an event in a terminal FAILED state:
For contrast,
gateway event retryon the same event genuinely worked — attempts went 1 → 2.Assessment
This is plausibly a legitimate server-side no-op: a terminal event cannot be cancelled or muted. The defect is that the CLI asserts it happened.
Event <id> muted.is a statement about state that is not true.Expected
Either report that the event was not in a state where the action applies, or report what the API actually did. A confirmation message should not be printed unconditionally on a 2xx that changed nothing.
Scope
Pre-existing, not a regression. Verified identical on
fbb924aand the merged branch.Why it matters
Same shape as several defects fixed in #392: the command reports success while nothing happened. Someone muting a noisy event believes it is muted.
Filed by Claude on Phil's behalf, from release-candidate regression testing.