Use the RequestHunt MCP server command

Before you start

Use this guide when you want an AI-assisted research workflow to connect to RequestHunt through the command listed by ReScience Lab.

You need:

• a Node-compatible environment that can run npx
• network access that allows package execution
• an MCP-compatible client if you want to register the server with an agent
• the current client-specific MCP configuration format for your agent or desktop app

This page documents the public command ReScience Lab lists for RequestHunt. It does not invent extra flags, environment variables, package internals, or data coverage claims.

Before connecting it to a long-running agent workflow, decide what kind of evidence you want back: repeated feature requests, complaint patterns, exact market language, or examples that can support a PRD. That makes the agent prompt easier to evaluate and easier to reuse later.

Steps

1. Confirm that npx is available:

   npx --version

2. Run the RequestHunt MCP server command:

   npx -y @resciencelab/requesthunt-mcp-server

3. If your MCP client requires an explicit command registration, add that command using the client’s current configuration format.

4. Start a research workflow that asks the agent to use RequestHunt for feature-request discovery, repeated complaints, or market language from public communities.

5. Keep downstream claims scoped. Treat RequestHunt outputs as research inputs that still need human review before becoming product requirements, comparison copy, or public claims.

Verify it worked

A working setup should let the MCP-compatible client see or call the RequestHunt server. The exact UI depends on the client. In a useful research run, the agent should be able to use RequestHunt context before drafting a brief, PRD, use-case page, or Pagewell content request.

Use the research safely

Treat RequestHunt output as evidence to review, not as final market truth. A strong agent workflow should preserve examples, repeated wording, and the source context behind each opportunity. Before turning research into public copy, check whether the examples are representative, current, and relevant to your target audience.

For Pagewell workflows, a useful handoff is a short brief: target audience, repeated pain point, source examples, search intent, claim limits, and the page type that should be generated. For Hal workflows, convert the research into a product requirements document with explicit user stories, non-goals, and acceptance criteria. That keeps discovery, writing, and implementation connected without overstating what the research proves.

Common issues

• If npx is unavailable, install or repair Node.js/npm in your environment.
• If package execution is blocked, check local security policy, firewall rules, or network access.
• If your MCP client does not detect the server, review the client’s MCP configuration schema and restart the client if required.
• If an agent overstates research results, narrow the prompt and require citations or examples before accepting downstream copy.

Related pages

• RequestHunt product page
• Find feature requests from Reddit
• MCP server definition
• Feature request pattern analyzer

Sources

• RequestHunt
• Model Context Protocol documentation
• npm npx documentation
• Last checked: 2026-05-09