---
name: libllm
description: >
  libllm - LLM API client for OpenAI-compatible endpoints. LlmApi class handles
  chat completions and embeddings via HTTP. Supports GitHub Models, Azure
  OpenAI, and standard OpenAI endpoints. Handles streaming responses, token
  counting, and multi-tool parallel call fixes. Use for LLM completions,
  embeddings, and AI model integration.
---

# libllm Skill

## When to Use

- Making chat completion requests to LLM providers
- Generating text embeddings for vector search
- Integrating with OpenAI-compatible APIs
- Handling streaming LLM responses

## Key Concepts

**LlmApi**: HTTP client for OpenAI-compatible endpoints. Handles authentication,
streaming, and response parsing.

**DEFAULT_MAX_TOKENS**: Standard token limit for completions.

## Usage Patterns

### Pattern 1: Chat completion

```javascript
import { LlmApi } from "@copilot-ld/libllm";

const api = new LlmApi(config, logger);
const response = await api.completion([{ role: "user", content: "Hello" }], {
  model: "gpt-4",
  maxTokens: 1000,
});
```

### Pattern 2: Generate embeddings

```javascript
const embeddings = await api.embed(["text to embed"]);
// Returns array of vectors
```

## Integration

Used by LLM service. Configurable via environment for different providers
(OpenAI, Azure, GitHub Models).