Fix: Prioritize task output_json over LLM response_format #3640
+204
−3
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
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
This commit fixes issue #3639 by ensuring task-level output settings (output_json and output_pydantic) take precedence over agent-level LLM response_format when both are set with Pydantic models. Changes: - Modified LLM._prepare_completion_params() to accept from_task parameter and check if task has output_json or output_pydantic set - If task has output settings, LLM's response_format is ignored - Updated LLM.call() to pass from_task to _prepare_completion_params() - Added comprehensive test to verify the priority behavior The fix ensures predictable behavior following the standard configuration hierarchy where more specific (task-level) settings override general (agent-level) defaults. Co-Authored-By: João <[email protected]>
🤖 Devin AI EngineerI'll be helping with this pull request! Here's what you should know: ✅ I will automatically:
Note: I can only respond to comments from users who have write access to this repository. ⚙️ Control Options:
|
…e_format Applied ruff auto-fixes to remove trailing whitespace from docstring and blank lines in the new test function. Co-Authored-By: João <[email protected]>
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Outdated suggestions cannot be applied.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.
Fix: Prioritize task output_json over LLM response_format
Summary
This PR fixes issue #3639 by establishing a clear priority hierarchy when both a task's
output_json/output_pydanticand an agent's LLMresponse_formatare set with Pydantic models. Task-level settings now take precedence over agent-level settings, following standard configuration hierarchy principles.Key changes:
LLM._prepare_completion_params()to accept afrom_taskparameter and check for task output settingsoutput_jsonoroutput_pydanticset, the LLM'sresponse_formatis ignoredLLM.call()to pass the task object through the call chainReview & Testing Checklist for Human
task.output_json=TaskModelandagent.llm.response_format=LLMModeland verify the task model is used in the resultoutput_jsonandoutput_pydanticfunctionality still works as expected when no LLMresponse_formatis setNotes
Requested by: João ([email protected])
Link to Devin run: https://app.devin.ai/sessions/b30fa7d430ca45ee99cd2f492b080702