ChatGPT / OpenAI Integration

ChatGPT supports MCP servers natively through the ChatGPT Desktop app. Connect Snipara so ChatGPT can query your documentation directly during conversations.

What You Get

ChatGPT Desktop App (Recommended)

The ChatGPT Desktop app for macOS and Windows supports MCP servers natively. This is the easiest way to connect Snipara to ChatGPT.

Prerequisites

• ChatGPT Desktop app (macOS or Windows)
• A ChatGPT Plus, Team, or Enterprise subscription
• A Snipara project with at least one document indexed
• An API key from your Snipara dashboard

Step 1: Open MCP Settings

In the ChatGPT Desktop app, open Settings (gear icon) and navigate to the MCP Servers section under Beta features.

Step 2: Add Snipara Server

Click Add Server and enter the following configuration:

Step 3: Enable the Server

After adding the server, toggle it ON. ChatGPT will connect and discover Snipara's tools automatically. You should see a green indicator once connected.

Step 4: Start Querying

Open a new chat and ask questions about your documentation. ChatGPT will automatically use Snipara tools when relevant:

How does authentication work in this project?

[Using snipara.rlm_context_query]

Based on your documentation, authentication works by...

Alternative: stdio Package

If the ChatGPT Desktop app supports stdio-based MCP servers in your version, you can use the snipara-mcp PyPI package:

pip install snipara-mcp

or run without installing:

uvx snipara-mcp

Then configure in the ChatGPT Desktop MCP settings:

{

  "mcpServers": {

    "snipara": {

      "command": "uvx",

      "args": ["snipara-mcp"],

      "env": {

        "SNIPARA_API_KEY": "YOUR_API_KEY",

        "SNIPARA_PROJECT_ID": "YOUR_PROJECT_SLUG"

      }

    }

  }

}

OpenAI API Integration

You can also use Snipara with the OpenAI API directly. Query Snipara's REST API or MCP endpoint, then pass the retrieved context to your OpenAI completions.

import openai

import requests

# 1. Query Snipara for relevant context

resp = requests.post(

    "https://api.snipara.com/mcp/YOUR_PROJECT_SLUG/tools/invoke",

    headers={"X-API-Key": "YOUR_API_KEY"},

    json={

        "name": "rlm_context_query",

        "arguments": {"query": "How does auth work?", "max_tokens": 4000}

    }

)

context = resp.json()

# 2. Pass context to OpenAI

client = openai.OpenAI()

completion = client.chat.completions.create(

    model="gpt-4o",

    messages=[

        {"role": "system", "content": f"Use this context:\n{context}"},

        {"role": "user", "content": "How does auth work?"}

    ]

)

This approach works with any OpenAI model (GPT-4o, GPT-4o mini, o1, etc.) and gives you full control over how context is injected into your prompts.

Custom GPTs

You can add Snipara as an Action in a Custom GPT so it can query your documentation on behalf of users.

Setup

1. Open the GPT Builder and go to the Configure tab
2. Scroll down to Actions and click Create new action
3. Set the Authentication to API Key with header X-API-Key
4. Import the OpenAPI schema from your Snipara project's API settings
5. Save and test with a documentation query

Available Tools

Once connected, ChatGPT can use all Snipara MCP tools. Here are the most commonly used:

See the MCP Tools Reference for a full list of all available tools with parameters and examples.

Troubleshooting

Next Steps