Description
I'm using Kilo Code Review integrated with GitHub, where a review is automatically triggered whenever a new Pull Request is opened.
In some cases, the AI model fails during the review process. However, even when the review does not complete successfully, Kilo still reports the GitHub check as successful and displays the message:
Code review completed successfully.
For example, in one of the failed runs, the session log shows the following error:
The model hit its output limit while reasoning and produced no actionable output.
Despite this error, the review status is still marked as Complete in Kilo and succeeded in GitHub.
Expected behavior
If the AI model fails, reaches an output/token limit, or is otherwise unable to produce a valid review, the GitHub check should not be reported as successful.
Ideally, the check should:
- Be marked as failed or incomplete.
- Clearly indicate that the review could not be completed.
- Optionally include the reason for the failure.
Actual behavior
The AI model fails and produces no actionable review, but Kilo reports the review as successfully completed in the Pull Request.
This can be misleading because developers may assume that the PR was successfully reviewed when, in reality, no review was produced.
Additional context
The attached screenshots show an example where the internal session log reports a model failure, while the GitHub integration reports the review as successful.
Description
I'm using Kilo Code Review integrated with GitHub, where a review is automatically triggered whenever a new Pull Request is opened.
In some cases, the AI model fails during the review process. However, even when the review does not complete successfully, Kilo still reports the GitHub check as successful and displays the message:
For example, in one of the failed runs, the session log shows the following error:
Despite this error, the review status is still marked as Complete in Kilo and succeeded in GitHub.
Expected behavior
If the AI model fails, reaches an output/token limit, or is otherwise unable to produce a valid review, the GitHub check should not be reported as successful.
Ideally, the check should:
Actual behavior
The AI model fails and produces no actionable review, but Kilo reports the review as successfully completed in the Pull Request.
This can be misleading because developers may assume that the PR was successfully reviewed when, in reality, no review was produced.
Additional context
The attached screenshots show an example where the internal session log reports a model failure, while the GitHub integration reports the review as successful.