Skip to content
MCP server

mockcrowd

Design AI personas from real posts and test whether an idea has demand

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted, Linux
Runs
Locally
Version
0.4.2
Listing updated
today

Similar MCP servers

About mockcrowd

mockcrowd is an open-source MCP server for analytics and data. It runs locally on your machine and is started by your AI client.

Most persona generators invent each person one trait at a time. mockcrowd instead scores real posts from Hacker News, Bluesky, Reddit, GitHub and optionally the App Store on traits such as budget sensitivity, vendor trust and pain urgency, measures how those traits move together, then samples personas that follow those links. Each persona cites the three real posts behind it, and each score is marked stated or inferred.

The population can then vote blind against a bland control pitch, so the verdict is the lift over the control rather than a raw yes rate. A simulated Reddit-style discussion of about 20 personas follows the OASIS design. The README shows a 20-persona run where 60% would try an idea free and 5% would pay alone, falling to 35% and none after discussion.

It runs locally from PyPI (Python 3.10+) over stdio, using Claude Code, Codex, Code Puppy, Ollama or any OpenAI-compatible API. Only the Claude adapter has been run live end to end, per the README. Reddit needs a free API key; the other sources need none.

Install

Run in a terminal

claude mcp add mockcrowd -- uvx mockcrowd

Add to claude_desktop_config.json

{
  "mcpServers": {
    "mockcrowd": {
      "command": "uvx",
      "args": [
        "mockcrowd"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "mockcrowd": {
      "command": "uvx",
      "args": [
        "mockcrowd"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "mockcrowd": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "mockcrowd"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "mockcrowd": {
      "command": "uvx",
      "args": [
        "mockcrowd"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "mockcrowd": {
      "command": "uvx",
      "args": [
        "mockcrowd"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

What it does

Trait-linked personas
Samples personas from measured correlations between traits.
Cited sources
Every persona lists the real posts it was built from.
Blind vote with control
Compares interest against a bland control pitch to counter agreeable models.
Simulated discussion
Runs a Reddit-style thread with comment, reply and vote actions.
Pinned traits
Fixes a trait such as vendor trust and adjusts linked traits.
Choice of model
Works with Claude Code, Codex, Code Puppy, Ollama or OpenAI-compatible endpoints.

Good for

  • Check demand for an app idea before building it
  • Estimate who would pay and who would refuse
  • Backtest on a real feature request
  • Design a persona set for an audience such as forgetful subscribers

Pricing

Open source

mockcrowd FAQ

What do I need to run it?

Python 3.10 or newer and pip install mockcrowd, plus one supported AI such as Claude Code, Codex, Ollama or an OpenAI-compatible API.

Does it need API keys for sources?

Only Reddit, which needs a free script app. Hacker News, Bluesky and GitHub need none.

How many LLM calls does a run use?

The README says a 500-persona run makes roughly 100 to 150 calls.

Which AI backends are verified?

Claude has been run live end to end; the other adapters are tested against their documented flags and APIs but not yet in a live run.

Is mockcrowd free?

Yes. mockcrowd is open source, so you can use and self-host it for free.

How do I add mockcrowd to Claude Code?

Run this command in your terminal: claude mcp add mockcrowd -- uvx mockcrowd

What are the best alternatives to mockcrowd?

The closest alternatives to mockcrowd are headroom, omniget, next-ai-draw-io, mcp-for-blender and Agent-Native Analytics.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers