fix: preserve tool_calls in OpenAI input messages #335
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Add helper function to format tool calls from both API responses and input messages - Handle multimodal content and tool calls in structured format - Maintain backwards compatibility for simple string content - Add test coverage for messages with tool calls
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
Comment on lines
1245
to
1249
| # Third message should be the tool response | ||
| assert props["$ai_input"][2] == { | ||
| "role": "tool", | ||
| "content": "72 degrees and sunny", | ||
| } |
There was a problem hiding this comment.
style: Missing tool_call_id field validation in the tool message assertion - should verify this field is preserved
Suggested change
| # Third message should be the tool response | |
| assert props["$ai_input"][2] == { | |
| "role": "tool", | |
| "content": "72 degrees and sunny", | |
| } | |
| # Third message should be the tool response | |
| assert props["$ai_input"][2] == { | |
| "role": "tool", | |
| "tool_call_id": "call_abc123", | |
| "content": "72 degrees and sunny", | |
| } |
Prompt To Fix With AI
This is a comment left during a code review.
Path: posthog/test/ai/openai/test_openai.py
Line: 1245:1249
Comment:
**style:** Missing `tool_call_id` field validation in the tool message assertion - should verify this field is preserved
```suggestion
# Third message should be the tool response
assert props["$ai_input"][2] == {
"role": "tool",
"tool_call_id": "call_abc123",
"content": "72 degrees and sunny",
}
```
How can I resolve this? If you propose a fix, please make it concise.
- Extract and preserve tool_call_id from input messages - Keep simple string content for tool messages (with optional tool_call_id) - Update test to verify tool_call_id is preserved for tool responses 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <[email protected]>
Radu-Raicea
reviewed
Oct 6, 2025
| "content": content_value, | ||
| } | ||
| # Preserve tool_call_id for tool role messages | ||
| if tool_call_id: |
There was a problem hiding this comment.
Can we have a tool_call_id if we don't have any tool_calls?
Radu-Raicea
reviewed
Oct 6, 2025
| "content": content_items, | ||
| } | ||
| ) | ||
|
|
There was a problem hiding this comment.
Do we need to change anything for the responses API format?
Radu-Raicea
reviewed
Oct 6, 2025
| } | ||
| ) | ||
| content.extend( | ||
| _format_tool_calls_to_content(choice.message.tool_calls) |
There was a problem hiding this comment.
This fixes it for the chat completions response API format. What about the responses API format?
Radu-Raicea
reviewed
Oct 6, 2025
| assert props["$ai_tools"] == tools | ||
|
|
||
|
|
||
| def test_input_messages_with_tool_calls(mock_client, mock_openai_response): |
There was a problem hiding this comment.
Do you think that there are other tests that are testing things with tool_calls? Maybe we can just put some extra assertions there?
We should probably also test some of the edge cases that you have in your code
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem
Tool calls in input messages were being dropped during formatting, causing loss of conversation context when passing multi-turn conversations with tool usage to OpenAI's API.
Solution
Added proper handling to preserve
tool_callsin input messages by:_format_tool_calls_to_content()helper function that handles both object-based tool calls (from API responses) and dict-based tool calls (from user input)format_openai_input()to detect and preserve tool calls alongside text contentChanges
Testing
test_input_messages_with_tool_calls()to verify tool call preservation in multi-turn conversationsTested locally by sending some events using the fix branch of
posthog-python:Previously the tool calls were being dropped in the assistant input messages that used
tool_calls.(slack thread here for further context/background)