Skip to main content

LLM-MCP (LLM Platform Model Context Protocol)

LLM-MCP is the orchestration and agent registry server for the Acme AI Platform. It provides a unified protocol for tool registration, vector operations, and model management. This documentation hub contains comprehensive guides, references, and examples for working with the LLM-MCP system.

Table of Contents​

Documentation Standards​

Quick Start​

Get started with LLM-MCP in minutes:

  1. Install dependencies: 
    npm install
  2. Start the server: 
    npm run start
  3. Access the OpenAPI explorer:

Architecture​

LLM-MCP integrates with other platform components through REST APIs and standard protocols:

graph TD;
  A[Drupal llm_platform_mcp UI] -->|OpenAPI, REST| B(llm-mcp Server);
  B -->|OpenAPI, REST| C(GitLab CI/CD);
  C -->|Status API| D[Drupal UI: Deploy Status];

Key architectural components:

  • Agent actions in Drupal trigger llm-mcp endpoints and GitLab pipelines
  • OpenAPI contracts are validated in CI/CD
  • Pipeline/job status is surfaced in the UI

Documentation Sections​

The documentation is organized into these main sections:

API & Usage​

Architecture & Design​

Guides & Setup​

Operations & Deployment​

Reference​

Common Use Cases​

LLM-MCP supports these common use cases:

  • Tool Registration and Discovery: Register custom tools and make them available for AI agents
  • Vector Operations: Store, retrieve, and search vector embeddings for semantic search
  • AI Streaming: Real-time AI response streaming with metrics tracking
  • Function Calling: Register and execute functions with AI models
  • Retrieval Augmented Generation (RAG): Implement RAG workflows with vector operations
  • Multi-Provider Support: Abstract multiple AI providers behind a unified interface
  • Agent Orchestration: Coordinate multiple AI agents and tools
  • GitLab CI/CD Integration: Automate deployment and validation through pipelines
  • Drupal Integration: Connect LLM-MCP with Drupal-based UIs

Contributing​

To contribute to LLM-MCP documentation:

References​

For questions or contributions, see the Acme AI Platform documentation hub or contact the platform team.- DIRECT_MCP_TOOL_EXECUTION.md