🧩 Supercharge ChatGPT with a Custom MCP Server: AVR Wiki Integration (Wiki.js)

One of the most underrated new features in ChatGPT is the ability to connect it to your own MCP (Model Context Protocol) servers.

This means ChatGPT can pull answers directly from your documentation or APIs — no more endless scrolling through wikis.

If, like me, you’re a bit lazy when it comes to reading docs 😉, this is a game-changer.

🚀 Why MCP Servers?

Normally, ChatGPT answers using its training data and some browsing. With MCP, you can:

• 📖 Access your private or specialized documentation directly in ChatGPT
• ⚡ Save time by asking instead of searching
• 🧑‍💻 Query APIs or tools from within the chat
• 🔗 Always stay up-to-date with the latest content from your own team

Imagine asking:

“How can I configure the docker-compose file to work with Agent Voice Response and OpenAI Realtime?”

And ChatGPT replies with the exact snippet from your docs.

🌐 AVR Wiki MCP Server

We’ve exposed our Agent Voice Response (AVR) Wiki MCP server at:

👉 https://wikimcp.agentvoiceresponse.com/mcp

This means you can hook ChatGPT directly into the AVR Wiki and ask it about:

• 🔊 Integrating Deepgram ASR or Google Speech
• 🤖 Running Ollama locally as your LLM
• 📞 Connecting AVR with Asterisk, FreePBX, or VitalPBX
• 🗣️ Using OpenAI Realtime or Gemini STS for speech-to-speech

⚙️ How to Set It Up

1. Open ChatGPT → Settings → Beta Features
2. Enable Custom GPTs and MCP Servers
3. Go to Settings → MCP Servers
4. Add a new server:
5. Name: AVR Wiki MCP
6. Endpoint: https://wikimcp.agentvoiceresponse.com/mcp
7. Save, start a new chat, and you’re ready to roll 🎉

🧪 Example in Action

After setup, you can ask:

“Show me an example docker-compose with Deepgram ASR and Anthropic Claude LLM.”

or

“What are the environment variables for Gemini STS?”

ChatGPT will answer using real data from the AVR Wiki, no manual searching needed.

✅ Why This Matters

• 🕒 Faster onboarding → new teammates can ask ChatGPT instead of reading docs line by line
• 🐍 Developer productivity → spend less time digging, more time building
• 🧠 Always accurate → responses come straight from the official docs

With MCP, your documentation stops being a static site and becomes an interactive assistant inside ChatGPT.

For us at Agent Voice Response, this is a huge step forward in making AI tools even more developer-friendly.

👉 Try it out today: https://wikimcp.agentvoiceresponse.com/mcp