e3507a7f6c
**Critical Fix for LiteLLM Integration** Changed the default model provider from "openai" to "litellm" when no provider is explicitly configured. This fixes the 400 Bad Request error with prompt_cache_key field. ## The Problem - Default provider was hardcoded to "openai" (line 983) - OpenAI provider uses wire_api = Responses API - Responses API sends prompt_cache_key field - LiteLLM/Anthropic Chat Completions API rejects this field → 400 error ## The Solution - Changed default from "openai" to "litellm" - LiteLLM provider uses wire_api = Chat API - Chat API does NOT send prompt_cache_key - Now works with LiteLLM proxy out of the box ## Impact Users can now run LLMX without any config and it will: - Default to LiteLLM provider - Use anthropic/claude-sonnet-4-20250514 model - Connect to $LITELLM_BASE_URL or http://localhost:4000/v1 - Use Chat Completions API (correct for LiteLLM) 🤖 Generated with Claude Code Co-Authored-By: Claude <noreply@anthropic.com>
llmx-core
This crate implements the business logic for LLMX. It is designed to be used by the various LLMX UIs written in Rust.
Dependencies
Note that llmx-core makes some assumptions about certain helper utilities being available in the environment. Currently, this support matrix is:
macOS
Expects /usr/bin/sandbox-exec to be present.
Linux
Expects the binary containing llmx-core to run the equivalent of llmx sandbox linux (legacy alias: llmx debug landlock) when arg0 is llmx-linux-sandbox. See the llmx-arg0 crate for details.
All Platforms
Expects the binary containing llmx-core to simulate the virtual apply_patch CLI when arg1 is --llmx-run-as-apply-patch. See the llmx-arg0 crate for details.