llms-full MCP Server

Example Use Cases

• 1

  Automatically discover and crawl web documentation topics and convert them into a local, searchable knowledge base.

• 2

  Synthesize and index mixed local files and web pages into a Qdrant vector store for fast RAG-style retrieval.

• 3

  Task-level control and observability (status, cancel, restart) when running multi-stage content ingestion pipelines with resource-aware concurrency.

Description

This MCP server provides a unified, sequential pipeline (Discovery → Fetch → Synthesize → Embed → Cleanup) to discover web or local sources, extract content, generate LLM summaries, embed them, and index into a Qdrant vector store. It integrates Tavily for web search/discovery and supports multiple LLM and embedding providers (Gemini, Ollama, OpenRouter, Chutes, OpenAI, Google). The project includes task management (status, cancel, restart), concurrency controls for browser/LLM/Qdrant operations, and utilities for vector store management and RAG-style search. It simplifies creating and maintaining a local searchable documentation knowledge base for agents and teams.