JitAPI: Just-in-Time API Orchestration for LLMs
JitAPI is an advanced MCP server designed for seamless interaction between large language models (LLMs), such as Claude, and various APIs. By dynamically discovering relevant endpoints from OpenAPI specifications, JitAPI streamlines the API interaction process without the need to load entire API specifications into memory. Instead, it leverages semantic search and dependency graphs to pinpoint only the endpoints essential for specific tasks.

Key Features

• Semantic Search: Queries can be conducted in natural language to efficiently find relevant API endpoints.
• Dependency Graphs: Automatically identifies endpoint dependencies to ensure the required parameters are passed correctly, enhancing workflow efficiency.
• LLM-Powered Workflow Planning: Uses GPT-4o-mini to intelligently extract parameters from user queries and formulate multi-step workflows.
• Generic Execution: Facilitates the execution of workflows with automatic parameter passing between various steps.
• Multi-API Support: Enables simultaneous registration and querying of multiple APIs, enhancing usability.
• MCP Integration: Provides native integration with Claude Code via the Model Context Protocol (MCP).

Example Workflow

Utilizing JitAPI allows users to interact with APIs in a conversational manner. For instance:

User: "Register the Petstore API from https://petstore.swagger.io/v2/swagger.json"  
Claude: [calls jitapi:register_api tool]  
✓ Registered Swagger Petstore with 20 endpoints  
  
User: "Find a pet and get its details"  
Claude: [calls jitapi:get_workflow tool]  
Workflow planned:  
1. GET /pet/findByStatus - Find pets by status  
2. GET /pet/{petId} - Get pet details  
Parameters extracted: status="available" (from context)  
  
User: "Execute that workflow"  
Claude: [calls jitapi:set_api_auth, then jitapi:execute_workflow]  
Step 1: Found 3 available pets  
Step 2: Retrieved details for pet "Max"  

Architecture Overview

JitAPI employs a sophisticated architecture split into two primary pipelines:

1. Ingestion Pipeline: Responsible for parsing OpenAPI specifications, building graphs, and storing them.
2. Runtime Pipeline: Handles user queries, performs vector searches, and executes workflows based on the planned steps.

Configuration Requirements

To function effectively, JitAPI necessitates an OpenAI API key for embeddings and workflow planning. This key can be set through various configuration options including MCP settings, environment variables, or a .env file located in specified directories.

Supported Authentication Methods

JitAPI supports multiple authentication methods such as API Key in headers, query strings, and Bearer tokens to ensure secure API calls.