Fix Cohere streaming and xAI tools validation

[jxnl]

Summary

• use cohere chat_stream for streaming and handle async return types safely
• add Cohere streaming chunk extraction for Partial/Iterable
• validate xAI tools responses with prepared models and unwrap adapter results

Testing

• not run (API keys required)

---

Important

Enhance Cohere streaming and xAI tools validation by using chat_stream and validating responses with prepared models.

• Cohere Streaming:
  • Use chat_stream for streaming in from_cohere() in client.py.
  • Add chunk extraction logic for Mode.COHERE_TOOLS and Mode.COHERE_JSON_SCHEMA in extract_json() and extract_json_async() in iterable.py and partial.py.
• xAI Tools Validation:
  • Validate responses with prepared models in acreate() and create() in client.py.
  • Unwrap adapter results in _validate_model_from_json() in function_calls.py.
• Misc:
  • Update ty-check args in .pre-commit-config.yaml to ignore unresolved imports.

^{This description was created by for 159f121. You can customize this summary. It will automatically update as commits are pushed.}

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed everything up to 83dd71c in 1 minute and 16 seconds. Click for details.

• Reviewed 464 lines of code in 5 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/dsl/iterable.py:153

• Draft comment:
  The COHERE extraction logic (handling event_type and chunk_type) is duplicated in both sync and async methods. Consider refactoring this block into a helper function to adhere to DRY principles.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

2. instructor/providers/cohere/client.py:84

• Draft comment:
  In the async_wrapper, using call_kwargs.pop("stream", False) directly modifies the caller’s dictionary. Consider copying the kwargs before modification to prevent unintended side effects.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

3. instructor/providers/xai/client.py:137

• Draft comment:
  The logic for selecting and preparing the response model (using prepare_response_model for XAI_TOOLS or streaming mode) could benefit from an inline comment clarifying the decision criteria for future maintainers.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

4. instructor/providers/xai/client.py:218

• Draft comment:
  The error message when no tool calls and no text content are found includes the raw response. Ensure that exposing raw responses does not leak sensitive data in production.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

5. instructor/providers/xai/client.py:312

• Draft comment:
  The non-streaming branch in the create function duplicates JSON extraction logic from both text and tool_call paths. Consider refactoring this repeated logic into a dedicated helper to reduce duplication.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

Workflow ID: wflow_r5S1IlEJEP8WT8IW

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

[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	159f121	Commit Preview URL Branch Preview URL	Jan 04 2026, 04:45 PM

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed 159f121 in 57 seconds. Click for details.

• Reviewed 589 lines of code in 3 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. .pre-commit-config.yaml:37

• Draft comment:
  The ty-check hook now ignores unresolved imports. Please add a brief comment explaining why '--ignore unresolved-import' is needed.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

2. instructor/dsl/iterable.py:157

• Draft comment:
  There is repeated JSON start extraction logic (search for '{' or '[' and slicing) in multiple event branches. Consider refactoring this pattern into a helper function to adhere to the DRY principle.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

3. instructor/dsl/iterable.py:370

• Draft comment:
  The asynchronous JSON extraction (in extract_json_async) duplicates similar logic to its synchronous counterpart. Centralize the JSON start extraction into a shared helper to improve maintainability.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

4. instructor/dsl/partial.py:288

• Draft comment:
  Multiple branches in the extract_json function show nearly identical logic for locating the JSON start index. Refactor these repeated blocks into a common helper to reduce duplication.
• Reason this comment was not posted:
  Confidence changes required: 80% <= threshold 85% None

5. instructor/dsl/partial.py:245

• Draft comment:
  Potential bug: In model_from_chunks, 'chunk_buffer += chunk' is used, which concatenates the characters of the string instead of appending the full chunk. Use 'chunk_buffer.append(chunk)' to correctly accumulate whole chunks.
• Reason this comment was not posted:
  Comment was not on a location in the diff, so it can't be submitted as a review comment.

Workflow ID: wflow_KlNJhd3BDhfMG1M8

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