PolygraphStaging

How do you want to start?

Polygraph works inside your AI agent, on either side of a request.

How do you want to start?
  1. 1

    Copy the server address

    https://staging-mcp.usepolygraph.com/mcp

  2. 2

    Add it to your AI agent

    Your AI agent
    1. Open Plugins in ChatGPT, in the desktop app or on the web (chatgpt.com/plugins). Choose Add, then Create custom MCP server.
    2. Name it Polygraph Staging, paste the server address as the Server URL and keep Authentication on OAuth.
    3. Tick “I understand and want to continue”, then choose Create as a plugin.
    4. Choose Continue to Polygraph, sign in and choose Allow.
    1. Open Settings in Claude, then Connectors.
    2. Add a custom connector named Polygraph Staging and paste the server address.
    3. Sign in to Polygraph.

    Sign in with your email. Signing in doesn’t agree to anything.

    Sign in with the address the request was sent to. Signing in doesn’t agree to anything.

  3. 3

    Say what you want to know

    In your own words, for example:

    “Ask Rodwood how many enterprise retailers use their software.”

    Polygraph turns it into a few short questions. Nothing is sent until you approve them.

    Ask for the request

    “Show my Polygraph requests.”

    Reading it agrees to nothing. You can ask about it, agree, suggest changes or decline.

You get short, verified answers, all together and dated. The person you ask looks through their own records with their own AI agent. They never hand them over.

If you agree, your own AI agent looks through your records privately and shows you its answers first. Nothing is sent until you approve them, and then they go together. Not every AI agent can send answers yet. When you open a request, yours tells you whether it can.

Polygraph never looks through anyone’s records. The person asked uses their own AI agent, and only the answers they approve come back, together and dated.

Questions? support@polygraphagent.com