FutureSearch MCP Server
Use FutureSearch’s cohort API from MCP to research, score, classify, or forecast every row of a dataset inside your agent.
Overview
FutureSearch MCP is an MCP server for the idea phase that connects agents to FutureSearch so they can research, score, classify, or forecast each dataset row.
What is this MCP server?
- PyPI package futuresearch-mcp (version 0.9.0) with stdio via uvx
- Same row-level verbs as Everyrow: forecast, score, classify, research
- Requires FUTURESEARCH_API_KEY from https://cohort.futuresearch.ai/app
- Branded FutureSearch MCP Server title in official schema
- Subfolder futuresearch-mcp in futuresearch/everyrow-sdk repository
- Package version 0.9.0 identifier futuresearch-mcp on PyPI
- stdio transport with uvx runtime hint
- One required secret: FUTURESEARCH_API_KEY
What problem does it solve?
You have a FutureSearch cohort workflow but your agent cannot invoke it until research operations are exposed as MCP tools with your API key.
Who is it for?
Builders already on FutureSearch cohort who want Claude Code or Cursor to drive tabular research without leaving the editor.
Skip if: Users without a FutureSearch account who only need occasional web search or who want the everyrow.io branded endpoint instead.
What do I get? / Deliverables
Once futuresearch-mcp is registered with FUTURESEARCH_API_KEY, row-level cohort research runs inside agent sessions alongside your repo work.
- Cohort-backed per-row research results inside agent workflows
- Scored or classified tables ready for validate-phase decisions
- Stdio MCP server block wired into your agent settings
Recommended MCP Servers
Journey fit
Market and opportunity research tables belong in idea before you freeze MVP scope or pricing. FutureSearch targets systematic row-level investigation—the same research subphase as competitive and audience discovery work.
How it compares
FutureSearch-branded sibling to Everyrow MCP—same row-research pattern, different API key and cohort app URL.
Common Questions / FAQ
Who is FutureSearch MCP for?
Indie builders and analysts using FutureSearch cohort who want MCP-native row research from Claude Code, Cursor, or Codex.
When should I use FutureSearch MCP?
Use it during idea-stage dataset work when you need forecasts, scores, or classifications per row before validation or build.
How do I add FutureSearch MCP to my agent?
Install futuresearch-mcp from PyPI with uvx stdio, set FUTURESEARCH_API_KEY from cohort.futuresearch.ai, and add the server to your MCP config.