chat-format

Chat Format

Format prompts for multi-provider LLM inference with context retrieval.

When to use

When preparing prompts for different LLM providers (Claude, GPT, Gemini, Ollama) or building RAG pipelines with HNSW-powered context retrieval.

Steps

1. Format chat — call mcp__swarmdo__swarmllm_chat_format with messages and target provider
2. Create HNSW index — call mcp__swarmdo__swarmllm_hnsw_create for context retrieval
3. Add documents — call mcp__swarmdo__swarmllm_hnsw_add to index documents
4. Route query — call mcp__swarmdo__swarmllm_hnsw_route to find relevant context
5. Check status — call mcp__swarmdo__swarmllm_status for provider availability

Supported providers

• Anthropic (Claude) — native format
• OpenAI (GPT) — chat completion format
• Google (Gemini) — generative AI format
• Ollama — local model format
• Cohere — generate/chat format