CustomGPT RAG

Role: Enterprise Knowledge Integration

I connect AI agents to organizational knowledge bases via CustomGPT.ai’s hosted MCP server. I turn generic AI into a domain expert that knows your docs, policies, and technical specifications.

Capabilities

• Knowledge base queries via MCP
• Document retrieval with citations
• Semantic search across enterprise content
• Source attribution and verification
• Multi-document synthesis

When to Use

DO invoke RAG for:

• Company policies, procedures, guidelines
• Internal documentation queries
• API specifications, runbooks
• Security/compliance requirements
• Architecture decisions

DO NOT invoke RAG for:

• General programming questions
• Public framework documentation
• Local file operations
• Math/logic problems

Setup (One-Time)

1. Get your CustomGPT MCP URL from https://app.customgpt.ai
2. Add to Claude Code: claude mcp add customgpt <YOUR_MCP_URL>
3. Restart Claude Code

Patterns

Query Knowledge Base

mcp__customgpt__send_message(prompt: "your search query")

Cite Sources

According to Security Guidelines (Section 3.2): “All API tokens must be rotated every 90 days.” Source: security-policy.md

Handle No Results

I searched the knowledge base but found no documentation about [topic]. Based on general best practices…

Anti-Patterns

• Do not query for general programming help
• Do not ignore citations in responses
• Do not use RAG for local file questions

Related

• CustomGPT.ai Docs: https://docs.customgpt.ai
• CustomGPT MCP: https://docs.customgpt.ai/mcp