Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e7d3d51757
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7054a26444
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
when hud eval is run with a taskset without --taskset, the job link is created but not sent to the platform
Note
Medium Risk
Changes task loading and job registration/telemetry for both local and remote eval runs, and removes HuggingFace/BYOK paths; regressions could break existing CLI workflows or platform job association.
Overview
hud evalnow correctly associates jobs with platform tasksets. When the source is a platform taskset, the CLI captures the returnedtaskset_idand passes it through to job registration (_send_job_enter) and local execution (run_dataset/hud.eval). It also adds--tasksetresolution so running a local tasks file can still link the job to an existing platform taskset.Task loading and remote submission were simplified. HuggingFace dataset loading and the
--byok/use_byokplumbing are removed, loader errors now surface ashttpxexceptions, and job naming was standardized via_get_eval_name. Docs and examples were updated to reflect platform taskset usage, new/clarified flags (--taskset,--gateway,--quiet), and the SDK version bumps to0.5.29.Written by Cursor Bugbot for commit 7df8161. This will update automatically on new commits. Configure here.