Crate vtcode_llm 

vtcode-llm - LLM Provider Abstraction

Provides a unified interface for multiple LLM providers including Gemini, OpenAI, Anthropic, Meta AI, DeepSeek, and Ollama.

Re-exports

pub use capabilities::ProviderCapabilities;

pub use client::AnyClient;

pub use client::ProviderClientAdapter;

pub use config_adapter::AdapterEvent;

pub use config_adapter::AdapterHooks;

pub use config_adapter::AdapterHooksProvider;

pub use config_adapter::OwnedProviderConfig;

pub use config_adapter::as_factory_config;

pub use config_adapter::as_factory_config_with_hooks;

pub use factory_types::ProviderConfig;

pub use factory_types::infer_provider_from_model;

pub use model_resolver::DynamicModelMeta;

pub use model_resolver::DynamicModelRef;

pub use model_resolver::ModelAvailability;

pub use model_resolver::ModelResolver;

pub use model_resolver::ResolvedModel;

pub use provider::LLMStream;

pub use provider::LLMStreamEvent;

pub use provider_config_types::ProviderConfig as ProviderConfigData;

pub use tool_bridge::CorrelationStats;

pub use tool_bridge::IntentFulfillment;

pub use tool_bridge::MessageCorrelationTracker;

pub use tool_bridge::MessageToolCorrelation;

pub use tool_bridge::ToolExecution;

pub use tool_bridge::ToolIntent;

pub use tool_bridge::ToolIntentExtractor;

Modules

capabilities

Compatibility re-export for provider capability metadata.

client

config_adapter

Provider configuration traits decoupled from VT Code’s dot-config storage.

copilot

error_display

LLM error display utilities with enhanced ANSI color support

factory_types

Core provider configuration type.

http_client

Centralized HTTP client factory for LLM providers.

model_resolver

open_responses

Open Responses specification conformance layer.

provider

Universal LLM provider abstraction with API-specific role handling

provider_base

Base traits and utilities for LLM providers

provider_config_types

Core provider configuration type.

providers

reasoning_effort

Capability-driven reasoning validation before a provider request is sent.

rig_adapter

single_response

tool_bridge

Bridge between messages and tool executions

types

usage_cost

Provider-normalized token usage and shared raw/cache-aware USD estimates.

utils

Shared utilities for LLM request/response processing

Macros

delegate_components

Wire multiple component names to provider types for a context.

Structs

LLMResponse

Universal LLM response structure

Usage

Enums

BackendKind

FinishReason

LLMError

LLM error types with optional provider metadata