vtcode-llm 0.171.5

LLM provider abstraction, client implementations, and streaming for VT Code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

# vtcode-llm

LLM provider abstraction layer for VT Code, providing a unified interface for multiple LLM providers including Gemini,
OpenAI, Anthropic, Meta AI, DeepSeek, Ollama, and Copilot.

<!-- cargo-rdme start -->

## vtcode-llm - LLM Provider Abstraction

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

<!-- cargo-rdme end -->

## Modules

| Module             | Purpose                                         |
| ------------------ | ----------------------------------------------- |
| `capabilities`     | Provider capability detection and feature flags |
| `client`           | Client trait and adapter implementations        |
| `config_adapter`   | Configuration adapter for provider settings     |
| `copilot`          | GitHub Copilot integration (feature-gated)      |
| `error_display`    | User-friendly error message formatting          |
| `factory_types`    | Factory pattern types for client creation       |
| `http_client`      | HTTP client implementations                     |
| `model_resolver`   | Model ID resolution and capability matching     |
| `open_responses`   | OpenAI Responses API support                    |
| `optimized_client` | Optimized client for high-throughput scenarios  |
| `provider`         | Provider trait and core types                   |
| `providers`        | Provider-specific implementations               |
| `rig_adapter`      | Rig framework integration                       |
| `system_prompt`    | System prompt construction                      |
| `tool_bridge`      | Tool call bridging between LLM and local tools  |
| `types`            | Core LLM types (messages, responses, streaming) |
| `utils`            | Shared utilities                                |

## Features

- **Multi-provider support**: Unified interface for 6+ LLM providers
- **Streaming**: Real-time response streaming
- **Tool calling**: Structured tool call support
- **Copilot integration**: GitHub Copilot OAuth flow (feature-gated)
- **Model resolution**: Automatic model capability detection

## Dependencies

- `vtcode-commons` (HTTP, error types, model families, CompactStr)
- `vtcode-config` (provider config, timeouts, auth)
- `vtcode-utility-tool-specs` (apply_patch schemas)