Dogg World Check

What dogg-world-check does

The dogg-world-check MCP server supports a ChatGPT app called Raptr Agent Builder. The app can turn an idea or meeting transcript into a single-file agent, using ChatGPT's model to write the agent while the MCP service supplies the template and validation logic.

It also provides access to the public open agent registry, known as RAR. Agents can be searched and their source retrieved through MCP tools. The service can explain how to run an agent either within the chat or on a computer running macOS, Windows, or Linux.

The server operates around the existing Brainstem ecosystem rather than changing Brainstem itself. Its documented behavior is to read the public registry and direct users to the public installer.

How it works

The dogg-world-check MCP server is available at https://airaptr-api.azurewebsites.net/mcp. The endpoint uses streamable HTTP, is stateless, and returns JSON responses. A compatible MCP client can connect to the endpoint and call the exposed tools.

A typical workflow starts with get_agent_template, which returns the official agent template and its rules. After an agent is produced, check_agent evaluates the file against those rules. find_agents searches RAR, while get_agent_code returns the source of a selected registry entry. use_agent_here lets ChatGPT run an agent against data supplied in the conversation, and how_to_run_agent provides optional instructions for running it locally.

Tools and capabilities

• Retrieve the official single-file agent template and authoring rules.
• Check a completed agent file against the template rules.
• Search the public RAR agent registry.
• Retrieve source code for a registry agent.
• Run an agent in the current ChatGPT conversation.
• Get local execution guidance for Mac, Windows, and Linux.

Setup and configuration

The hosted endpoint can be used without documenting a local installation step in the supplied material. For development, the repository includes commands to rebuild the embedded template, start a local server on port 8787 with Wrangler, test the checker against RAR agents, run an MCP smoke test, and deploy through Cloudflare or Azure Functions.

No required environment variables or license information are specified in the supplied README excerpt. The project provides separate site, privacy, and terms pages at https://airaptr.github.io/.

Limitations and notes

The service depends on the public RAR registry and public installer references. The supplied material does not describe private registry access, authentication, persistent state, or modifications to Brainstem. Local execution is optional; the documented in-chat runner is provided through use_agent_here.