fix: improve trigger error handling, markdown rendering, and release workflow#633
Conversation
…workflow - Mark trigger executions as failed when agent is not found/expired, has no LLM model, or model is unavailable instead of silently skipping - Add safety checks before expunging execution/trigger objects from DB session - Fix markdown bold/italic regex to avoid matching underscores inside words - Update GitHub Models draft step to use openai/gpt-5.4
yaojin3616
force-pushed
the
feature/update-release-workflow-triggers
branch
from
a523c28 to
387536a
Compare
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a523c286c9
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| shell: bash | ||
| env: | ||
| MODELS_TOKEN: ${{ secrets.MODELS_TOKEN }} | ||
| GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} |
There was a problem hiding this comment.
When use_ai_notes is true, this step now authenticates GitHub Models with the workflow GITHUB_TOKEN, but the workflow permissions block only grants contents and pull-requests; GitHub Actions sets omitted permissions to none, and the GitHub Models quickstart/API docs require models: read for workflows calling the inference endpoint. As a result, the curl request will return an authorization error and fail every AI-notes release run unless models: read is added.
Useful? React with 👍 / 👎.
1 participant
Changes
Backend
Frontend
CI/CD