Background: I’m using Responses API with o3-deep-research model attached with MCP tool. The MCP server has two tools search and fetch a per the specification and it searches on the file data and returns the meta data and fetches the file content by taking id as input. I’m using OpenAI 2.6.0 library, the o3-deep-research model is deployed in Azure Foundry.
Problem: The response API status is becoming incomplete with the reason as content filter. The file content which is being searched and fetched by the MCP server do not contain any content which might have triggered the content filtration. For some inputs its working fine.
- I do have agent who prepares the plan for the research topic, thats being passed to the o3-deep-research model.
- The o3-deep-research model have been prompted to do research.
- How to check what triggered the content filter , responses API seems not to have access to the details ?
- Is there a way to log the intermediate research results, so that at least we may get a clue ?