8 views
-/https://github.com/berriai/litellm/issues/27655
GitHub ยท issue

#27655 Responses API โ†’ Chat Completions transformation silently passes through unsupported built-in tool types

  • State: open
  • Author: @Ram-Pavith
  • Labels: llm translation

## Problem

When LiteLLM converts an OpenAI **Responses API** request to the **Chat Completions API**, several built-in tool types that have no valid equivalent in the Chat Completions spec are silently passed through to the provider. This causes provider-side `400` errors such as:

> "tools[N].type:type is illegal"

### Affected tool types - `local_shell` โ€” Codex CLI built-in shell tool (no Chat Completions equivalent) - `file_search` โ€” OpenAI file search tool (handled separately by emulated handler, but still leaks through in transformation) - `image_generation` โ€” OpenAI image generation tool - `namespace` โ€” Codex CLI 0.130.0 tool namespace - `mcp` โ€” MCP server tools are Responses-API-specific - `code_interpreter` โ€” Has no native Chat Completions equivalent, but previously passed through raw - `computer_use` / `computer_use_preview` โ€” Passed through raw, causing schema validation failures on generic providers - Unknown / passthrough types (e.g. `tool_search`, `code_execution_20250825`, Vertex AI native tools) โ€” silently forwarded

### Repro

```python from litellm.responses.litellm_completion_transformation.transformation import ( LiteLLMCompletionResponsesConfig, )

# This sโ€ฆ

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 ยท 31d17h57m22s ago โ€” entered ยท #import:https:::github.com:berriai:litellm post #2915
The left issue requires changes across request translation rules, compatibility handling, and regression coverage for multiple tool categories and provider behaviors. The right issue is narrower, centered on packaging or installation configuration and validation of the deployment path.
#0 of 0 ยท 31d17h54m38s ago โ€” current ยท #import:https:::github.com:berriai:litellm post #2961
25104 has broader cross-layer scope, involving credential modeling, deployment propagation, proxy batch flows, multiple cloud providers, and compatibility testing; 27655 is comparatively localized to Responses-to-Chat request transformation and tool filtering.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search