GPT-5.5 Agent mode fails after terminal tool execution with “function_call was provided without its required reasoning item”

Before submitting your bug report

• I've tried finding an answer on the Continue docs site
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS: macOS
- Continue version: 2.0.0
- IDE version:
- Model:gpt-5.5
- config:
  
name: Main Config
version: 1.0.0
schema: v1
models:
  - name: GPT-5.5
    provider: openai
    model: gpt-5.5
    apiKey: ${OPENAI_API_KEY}
    roles:
      - chat
      - edit
      - apply

Description

When using GPT-5.5 in Agent mode, Continue successfully executes terminal commands, but fails immediately after the tool execution.

The terminal command completes successfully, however Continue displays:

Error handling model response

and the request fails with:

{
  "message":"Item 'fc_...' of type 'function_call'
was provided without its required 'reasoning' item:
'rs_...'.",
  "type":"invalid_request_error",
  "param":"input"
}

To reproduce

⸻

Reproduction

1. Configure Continue with OpenAI GPT-5.5.
2. Open a local git repository.
3. Ask Continue to analyze all *.hcl files.
4. Continue invokes the Terminal tool.
5. The terminal command completes successfully.
6. Continue crashes while processing the model response.

Example command executed:

from pathlib import Path
for p in Path(".").rglob("*.hcl"):
    print(p)

The command finishes correctly.

The failure happens after tool execution.

⸻

Additional information

This issue is reproducible.

Initially I suspected DNS or NFS latency, but I reproduced it with:

• local repository
• local filesystem (no NFS)
• successful OpenAI API connectivity
• successful terminal execution

So the failure appears unrelated to networking.

Log output

The OpenAI Responses API works correctly when called directly with curl.

⸻

Error

Item 'fc_...' of type 'function_call' was provided without its required 'reasoning' item:
'rs_...'.
type=invalid_request_error
param=input

Reproduce the failure with GPT-5.5 in Agent mode by asking Continue to analyze *.hcl files in a local repository. Trace the request handling after the Terminal tool completes and compare the generated request with the direct OpenAI Responses API behavior. Done means the agent continues after terminal execution without the missing reasoning-item error.