Re-exports§
pub use shared::TagStreamSanitizer;pub use reasoning::clean_reasoning_text;pub use reasoning::ReasoningBuffer;pub use reasoning::ReasoningSegment;pub use reasoning::extract_reasoning_trace;pub use reasoning::split_reasoning_from_text;pub use anthropic::AnthropicProvider;pub use copilot::CopilotProvider;pub use custom_provider::CustomProviderBackendRouter;pub use deepseek::DeepSeekProvider;pub use evolink::EvolinkProvider;pub use gemini::GeminiProvider;pub use huggingface::HuggingFaceProvider;pub use llamacpp::LlamaCppProvider;pub use lmstudio::LmStudioProvider;pub use merge_gateway::MergeGatewayProvider;pub use merge_gateway_catalog::MergeCatalogAvailability;pub use merge_gateway_catalog::MergeCatalogFilters;pub use merge_gateway_catalog::MergeCatalogModel;pub use merge_gateway_catalog::MergeCatalogServiceTier;pub use merge_gateway_catalog::MergeCatalogSnapshot;pub use merge_gateway_catalog::MergeGatewayCatalogClient;pub use meta::MetaProvider;pub use mimo::MiMoProvider;pub use minimax::MinimaxProvider;pub use mistral::MistralProvider;pub use moonshot::MoonshotProvider;pub use nvidia::NvidiaProvider;pub use ollama::OllamaProvider;pub use openai::OpenAIProvider;pub use opencode_go::OpenCodeGoProvider;pub use opencode_zen::OpenCodeZenProvider;pub use openresponses::OpenResponsesProvider;pub use openrouter::OpenRouterProvider;pub use poolside::PoolsideProvider;pub use qwen::QwenProvider;pub use stepfun::StepFunProvider;pub use streaming_progress::StreamingProgressBuilder;pub use streaming_progress::StreamingProgressCallback;pub use streaming_progress::StreamingProgressTracker;pub use vercel::VercelProvider;pub use xai::XAIProvider;pub use zai::ZAIProvider;
Modules§
- anthropic
- Anthropic Claude API provider implementation
- anthropic_
types - base
- Base trait and common implementations for LLM providers
- common
- Shared provider helpers, grouped by responsibility behind stable imports.
- copilot
- custom_
provider - deepseek
- error_
handling - Centralized error handling for LLM providers Eliminates duplicate error handling code across providers
- evolink
- gemini
- huggingface
- llamacpp
- llama.cpp provider: a local, OpenAI-compatible server managed by vtcode.
- lmstudio
- LM Studio provider implementation
- local_
readiness - Unified readiness pre-flight for local inference providers.
- local_
server - merge_
gateway - Merge Gateway provider.
- merge_
gateway_ catalog - Merge Gateway authenticated model catalog discovery.
- meta
- Official Meta AI OpenAI-compatible provider.
- mimo
- minimax
- mistral
- moonshot
- nvidia
- NVIDIA NIM OpenAI-compatible provider.
- ollama
- Ollama provider implementation
- openai
- OpenAI Provider Implementation
- opencode_
go - opencode_
zen - openresponses
- OpenResponses provider module.
- openrouter
- poolside
- qwen
- reasoning
- shared
- Shared provider wire helpers with stable facade imports.
- stepfun
- StepFun provider backed by the native Responses API
(
POST {base}/responses). - streaming_
progress - Provider-agnostic streaming timeout progress tracking
- tool_
format - Provider tool formatters.
- vercel
- xai
- zai
Constants§
- DEFAULT_
VEC_ CAPACITY - Default capacity hints for common collections
- ERROR_
DETECTION_ PATTERNS - Standard error patterns used for error detection across tools
- MAX_
SEARCH_ RESULTS - Context optimization constants following AGENTS.md guidelines
- OVERFLOW_
INDICATOR_ PREFIX