AI Agent: Chat with GitHub OpenAPI Specs (RAG with OpenAI & Pinecone)

Unlock Instant GitHub API Mastery with this AI Agent

This n8n workflow transforms into an intelligent AI Agent designed to help you and your team master the GitHub API. It fetches the official GitHub OpenAPI specification, indexes it into a Pinecone vector database, and then uses OpenAI's powerful language models to answer your questions about API endpoints, parameters, request/response structures, and more, all through a simple chat interface. This AI-driven automation streamlines API exploration and integration tasks, saving valuable development time.

Key Features & Benefits

• AI-Powered Q&A: Leverages OpenAI's advanced LLMs to understand natural language questions about the GitHub API.
• Contextual Accuracy with RAG: Implements Retrieval Augmented Generation (RAG) by fetching relevant information from the GitHub OpenAPI spec (indexed in Pinecone) to provide accurate, up-to-date answers.
• Efficient Knowledge Retrieval: Utilizes Pinecone vector database for fast and scalable searching of API documentation.
• Automated Spec Ingestion: Includes a process to fetch and prepare the latest GitHub OpenAPI specification for indexing.
• Interactive Chat Interface: Offers a user-friendly chat experience within n8n to query API details directly.
• Developer Productivity Boost: Significantly reduces the time developers, CTOs, and automation leads spend manually sifting through extensive API documentation.
• Enhanced API Understanding: Helps in quickly grasping complex API functionalities for faster development and integration cycles.

• An n8n instance (Cloud or self-hosted).
• OpenAI API Key with access to a suitable model (e.g., gpt-4o-mini, gpt-3.5-turbo, or gpt-4).
• Pinecone account and API Key.
• A Pinecone index created in your account (e.g., "n8n-demo", ensure it matches the configuration in the Pinecone nodes). The vector dimension should be compatible with OpenAI's text-embedding-ada-002 (1536 dimensions) or the embedding model you choose.

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Credential Setup:
   • Locate all OpenAI nodes: 'OpenAI Chat Model' (two instances), 'Generate Embeddings', and 'Generate User Query Embedding'. Configure each with your OpenAI API Key.
   • Locate the 'Pinecone Vector Store' (for insertion) and 'Pinecone Vector Store (Querying)' nodes. Configure both with your Pinecone API Key and Pinecone Environment.
4. Pinecone Index Configuration:
   • In both Pinecone nodes ('Pinecone Vector Store' and 'Pinecone Vector Store (Querying)'), verify the Pinecone Index parameter (default: "n8n-demo"). Ensure this matches an existing index in your Pinecone account. If not, create the index in Pinecone with the correct vector dimension (e.g., 1536 for OpenAI's text-embedding-ada-002) or update the node parameter to your existing index name.
5. Data Indexing (Run Once to Populate Vector Store):
   • The top branch of the workflow (starting with 'When clicking ‘Test workflow’') is for indexing. Execute this branch by clicking 'Test workflow' on the 'When clicking ‘Test workflow’' node.
   • This will trigger the 'HTTP Request' node to download the GitHub OpenAPI specification, which is then processed and embedded by 'Generate Embeddings' and finally inserted into your Pinecone index via the 'Pinecone Vector Store' node.
   • Verify successful data insertion in your Pinecone dashboard.
6. Configure Chat Agent Logic:
   • Review the system message in the 'AI Agent' node. You can customize it to alter the agent's persona or instructions.
   • Ensure the 'Vector Store Tool' node is correctly linked and configured. Its purpose is to connect the AI Agent to your Pinecone knowledge base.
7. Activate Chat Functionality:
   • The 'When chat message received' node serves as the entry point for user queries.
   • You can interact with this agent via the n8n chat interface (accessible when you open the workflow) or by embedding the chat in your applications.
   • Activate the entire workflow (toggle switch at the top right of the n8n canvas) to make the chat agent live and responsive.