How run-llama/llamacloud-mcp works
This archived package exposes LlamaCloud index retrieval and file extraction as dynamically registered MCP tools. An operator configures indexes, extraction agents, IDs, API key, and transport through the CLI; an MCP client invokes the resulting tools and receives retrieved or extracted data. The sampled implementation is concentrated in llamacloud_mcp/main.py; the README documents supported transports and broader client usage, but those client examples are not repository modules.
Main entry points: main.py MCP Server
Components 10
CLI and MCP
- CLI Bootstrap
llamacloud_mcp/main.py - MCP Server
llamacloud_mcp/main.py
Index Retrieval
- Index Tool
llamacloud_mcp/main.py - LlamaCloud Index
File Extraction
- Extract Tool
llamacloud_mcp/main.py - Agent Listing API
- File Upload API
- Extraction Job API
Other components
- Operator
- MCP Client
Connections 15
- Operator → CLI Bootstrap: configures
- CLI Bootstrap → MCP Server: registers tools
- CLI Bootstrap → MCP Server: starts transport
- MCP Client → MCP Server: invokes tools
- MCP Server → Index Tool: dispatches query
- MCP Server → Extract Tool: dispatches file
- Index Tool → LlamaCloud Index: retrieves nodes
- LlamaCloud Index → Index Tool: returns results
- Extract Tool → Agent Listing API: lists agents
- Extract Tool → File Upload API: uploads file
- Extract Tool → Extraction Job API: starts extraction
- Extraction Job API → Extract Tool: returns data
- Index Tool → MCP Server: returns response
- Extract Tool → MCP Server: returns response
- MCP Server → MCP Client: sends result