Responses API validation error

[jxnl]

fix: Prevent reaskresponsestools from crashing with reasoning items

Describe your changes

Adds a check for hasattr(tool_call, "arguments") in the reask_responses_tools function. This prevents the function from attempting to access the .arguments attribute on ResponseReasoningItem objects, which do not have it, thus resolving a bug where validation errors were not properly sent back to the LLM during retry attempts when using reasoning models.

Issue ticket number and link

567-251: #1957

Checklist before requesting a review

• I have performed a self-review of my code
• If it is a core feature, I have added thorough tests.
• If it is a core feature, I have added documentation.

---

Linear Issue: 567-251

 

---

Important

Fixes crash in reask_responses_tools by checking for arguments attribute, ensuring proper handling of reasoning items.

• Behavior:
  • Adds hasattr(tool_call, "arguments") check in reask_responses_tools to prevent crashes with ResponseReasoningItem objects.
  • Ensures validation errors are sent back to LLM during retries.
• Functions:
  • Adds _filter_responses_tool_calls() to filter tool calls in utils.py.
  • Adds _format_responses_tool_call_details() to format tool call details in utils.py.
• Tests:
  • Adds test_reask_responses_tools_skips_reasoning_items_and_includes_details in test_streaming_reask_bug.py to verify reasoning items are skipped and tool details are included.

^{This description was created by for 8ad4b0b. You can customize this summary. It will automatically update as commits are pushed.}

[cursor]

Cursor Agent can help with this pull request. Just @cursor in comments and I'll start working on changes in this branch.
_{Learn more about Cursor Agents}

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	instructor	8ad4b0b	Commit Preview URL Branch Preview URL	Jan 16 2026, 02:03 PM

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed everything up to 8ad4b0b in 1 minute and 24 seconds. Click for details.

• Reviewed 146 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 4 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. instructor/providers/openai/utils.py:31

• Draft comment:
  Consider using a list comprehension in _filter_responses_tool_calls for conciseness and adding a comment that explicitly skips items (e.g. reasoning items) that don’t have an 'arguments' attribute.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 85% None

2. instructor/providers/openai/utils.py:44

• Draft comment:
  The helper _format_responses_tool_call_details cleanly extracts details using getattr. Consider adding type annotations for the 'tool_call' parameter to improve clarity.
• Reason this comment was not posted:
  Confidence changes required: 20% <= threshold 85% None

3. instructor/providers/openai/utils.py:134

• Draft comment:
  Ensure that _filter_responses_tool_calls reliably filters out items without the 'arguments' attribute so that accessing tool_call.arguments won’t cause runtime errors. The current design appears correct, but a brief inline note could clarify this intent.
• Reason this comment was not posted:
  Confidence changes required: 20% <= threshold 85% None

4. tests/test_streaming_reask_bug.py:138

• Draft comment:
  The test 'test_reask_responses_tools_skips_reasoning_items_and_includes_details' effectively validates that reasoning items are skipped and that tool call details are appended. Consider adding an additional case with multiple valid tool calls to further strengthen coverage.
• Reason this comment was not posted:
  Confidence changes required: 20% <= threshold 85% None

Workflow ID: wflow_vKnvg29gv7mJkRg0

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}