Changeflow has a built-in Model Context Protocol (MCP) server. This lets you connect Changeflow to any AI tool that supports MCP, including Claude, ChatGPT, Codex, Cursor, and VS Code Copilot.
Once connected, your AI assistant can search your changes, list your sources, start monitoring new pages, and trigger checks - all through natural conversation.
Setup
You need two things: your API token (from Integrations) and the MCP endpoint URL:
https://changeflow.com/mcp
Claude
Works on claude.ai, Claude Desktop, and Claude Mobile (Pro plan or above).
- Go to Settings > Connectors
- Click Add custom connector
- Paste the MCP endpoint URL:
https://changeflow.com/mcp - When prompted for authentication, use your API token from the Integrations page
That's it. Changeflow will appear as a connector you can enable in any conversation.
ChatGPT
Custom MCP servers on ChatGPT require a Business, Enterprise, or Edu plan. An admin sets it up once, then everyone on the team can use it.
Admin setup:
- Go to Workspace Settings > Permissions & Roles and enable Developer Mode
- Go to Settings > Apps > Create
- Enter the MCP endpoint URL:
https://changeflow.com/mcp - Configure the auth header:
Authorization: Bearer YOUR_TOKEN - Publish the app to your workspace
Team members:
- Go to Settings > Apps
- Find Changeflow and click Connect
- Start using it in any conversation
Developer Tools
If you use AI-powered coding tools, here are the setup instructions for those.
Claude Code
claude mcp add changeflow https://changeflow.com/mcp \
--header "Authorization: Bearer YOUR_TOKEN"
Cursor
Create .cursor/mcp.json in your project root (or ~/.cursor/mcp.json for global):
{
"mcpServers": {
"changeflow": {
"url": "https://changeflow.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_TOKEN"
}
}
}
}
VS Code (Copilot)
Create .vscode/mcp.json in your workspace:
{
"servers": {
"changeflow": {
"url": "https://changeflow.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_TOKEN"
}
}
}
}
Codex
Add to ~/.codex/config.toml (or .codex/config.toml in your project):
[mcp_servers.changeflow]
url = "https://changeflow.com/mcp"
http_headers = { "Authorization" = "Bearer YOUR_TOKEN" }
Available Tools
Once connected, your AI assistant has access to these tools:
Reading Data
| Tool | What it does |
|---|---|
| list_sources | List all your monitored sources. Filter by keyword, tag, or status. |
| get_source | Get full details of a source including its 25 most recent changes. |
| get_changes | Search recent changes across all sources. Supports keyword search, date filtering, and type filtering. |
| get_change | Get the full details of a single change, including the AI-filtered changeset. |
Managing Sources
| Tool | What it does |
|---|---|
| create_source | Start monitoring a new URL. Requires the URL, a brief describing what to look for, and a check frequency. |
| check_source | Trigger an immediate check on a source, regardless of its schedule. |
| pause_source | Pause monitoring on a source. |
| resume_source | Resume a paused source. |
Example Conversations
Here are some things you can ask your AI assistant once Changeflow is connected:
- "What changed on my monitored sources this week?"
- "Search my changes for anything about pricing updates"
- "Show me the latest changes from sources tagged 'competitors'"
- "Start monitoring https://example.com/pricing for pricing changes, check daily"
- "Pause my FDA source for now"
- "Check my SEC source right now"
- "What are the details of change v_12345?"
Authentication
All requests to the MCP endpoint use your API token via the Authorization: Bearer header. This is the same token used for the REST API. You can find it on the Integrations page.
Rate limit: 100 requests per hour per token. Each conversation typically uses 3-5 requests, so this is plenty for normal use.
How It Works
The MCP server at changeflow.com/mcp speaks the Model Context Protocol - an open standard for connecting AI tools to external services. When your AI assistant needs to look up changes or manage sources, it sends a request to this endpoint and gets structured data back.
No plugins to install, no extensions to download. Just connect your AI tool and authenticate.