GenAI config labels loss

[jxnl]

fix(genai): preserve config labels from dict config

Describe your changes

This PR resolves an issue where GCP billing/monitoring labels and cached_content were discarded or incorrectly handled when the config parameter was provided as a dictionary in Google GenAI provider functions. This was blocking enterprise usage due to tracking issues.

The changes update update_genai_kwargs, handle_genai_structured_outputs, and handle_genai_tools to correctly merge labels, cached_content, automatic_function_calling, and thinking_config from both dictionary-style and object-style config parameters. This ensures all relevant configuration fields are preserved and applied as intended, and cached_content correctly suppresses system_instruction and tool definitions when active.

New regression tests have been added to verify label preservation and cached content behavior for dictionary-style configs.

Issue ticket number and link

Linear: 567-269
GitHub: #1759

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: 567-269

 

---

Important

Fixes GenAI config handling to preserve labels and cached_content when provided as a dictionary, ensuring correct merging of configuration fields.

• Behavior:
  • Fixes issue where labels and cached_content were discarded when config was a dictionary in update_genai_kwargs, handle_genai_structured_outputs, and handle_genai_tools.
  • Ensures labels, cached_content, automatic_function_calling, and thinking_config are merged from both dictionary and object-style config.
  • cached_content suppresses system_instruction and tool definitions when active.
• Tests:
  • Adds regression tests in test_genai_config_merging.py to verify label preservation and cached content behavior for dictionary-style configs.
  • Tests ensure thinking_config, automatic_function_calling, and labels are correctly extracted and do not override existing base config values.

^{This description was created by for eb5f85a. 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	eb5f85a	Commit Preview URL Branch Preview URL	Jan 16 2026, 05:28 PM

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed everything up to eb5f85a in 1 minute and 14 seconds. Click for details.

• Reviewed 221 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 5 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/gemini/utils.py:332

• Draft comment:
  Good change: the extraction of thinking_config now supports both dicts and objects. Consider abstracting this extraction (and similar extraction for labels and cached_content) into a helper to reduce duplication.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 85% None

2. instructor/providers/gemini/utils.py:355

• Draft comment:
  The merging logic for 'automatic_function_calling', 'labels', and 'cached_content' now properly handles both dict and object styles. Consider DRYing up the repeated extraction by using a helper function.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 85% None

3. instructor/providers/gemini/utils.py:890

• Draft comment:
  In handle_genai_structured_outputs, extraction of thinking_config and cached_content now handles dict inputs correctly. The updated comments are clear.
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 85% None

4. instructor/providers/gemini/utils.py:970

• Draft comment:
  Similarly, handle_genai_tools now correctly extracts config fields from dict or object. Consider refactoring this common logic into a shared helper to keep the code DRY.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 85% None

5. tests/test_genai_config_merging.py:402

• Draft comment:
  Excellent and comprehensive tests covering both object and dict configs, including labels, cached_content, and thinking_config. They clearly validate the merged configuration behavior (issue #1759).
• Reason this comment was not posted:
  Confidence changes required: 0% <= threshold 85% None

Workflow ID: wflow_SGnig1h19DTyKTZY

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