mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-06-28 04:30:15 -05:00
064d23a6f8
* responses: skip known unsupported Responses tool types from Codex CLI - Skip namespace, web_search, image_generation tools instead of HTTP 500 - Reject unknown non-function tool types with controlled error - Preserve function tool conversion logic unchanged Fixes Codex CLI 0.133.0 compatibility where it sends mixed tool types. * responses: harden codex compatibility coverage * responses: expose Codex model catalog metadata
38 lines
1.3 KiB
Gherkin
38 lines
1.3 KiB
Gherkin
@llama.cpp
|
|
@server
|
|
@codex
|
|
Feature: Codex CLI Responses API Compatibility
|
|
|
|
Background: Server startup
|
|
Given a server listening on localhost:8080
|
|
And a model file tinyllamas/stories260K.gguf from HF repo ggml-org/models
|
|
And a model file test-model.gguf
|
|
And a model alias tinyllama-2
|
|
And 42 as server seed
|
|
And 256 KV cache size
|
|
And 32 as batch size
|
|
And 2 slots
|
|
And 64 server max tokens to predict
|
|
And Jinja templating enabled
|
|
Then the server is starting
|
|
Then the server is healthy
|
|
|
|
Scenario: Responses API accepts mixed tool types from Codex
|
|
Given a model test
|
|
And an OAI compatible responses request with mixed Codex tool types
|
|
Then the mixed Codex tools response succeeds
|
|
|
|
Scenario: Models endpoint includes Codex model catalog metadata
|
|
Given the Codex model catalog is requested
|
|
Then the Codex model catalog is compatible
|
|
|
|
Scenario: Probe request with empty input and max_output_tokens=1 is accepted
|
|
Given a model test
|
|
And a probe responses request with empty input and max_output_tokens=1
|
|
Then the probe response is accepted
|
|
|
|
Scenario: previous_response_id returns a controlled error
|
|
Given a model test
|
|
And a responses request with invalid previous_response_id
|
|
Then the previous_response_id request returns an error
|