Skip to main content
OpenClaw is an agentic CLI runtime — a terminal-based environment where autonomous agents can invoke shell commands, call APIs, and drive workflows without human intervention. It pairs well with PickFu because PickFu’s survey API supports a full autonomous research loop: an agent can generate assets, create polls, wait for real human respondents, and analyze results inside a single skill.

​
Why OpenClaw + PickFu

PickFu’s survey API lets an agent drive a generate → test → refine cycle against real human feedback instead of AI guesswork. Because OpenClaw surfaces shell commands and MCP tools from inside an autonomous runtime, a skill can generate assets, publish a survey, and come back to analyze responses once they’re in — turning consumer research into a continuous loop instead of a manual workflow. OpenClaw can use PickFu two ways: through the MCP server for natural-language tool calls, or by invoking the CLI inside a skill for scripted workflows. Both paths share the same PickFu account and authentication, and you can use them together in a single skill.

​
Integration paths

​
MCP server

Wire the PickFu MCP server into OpenClaw with one command: 
openclaw mcp set pickfu '{"url":"https://mcp.pickfu.com/mcp"}'
Or add to openclaw.json manually: 
{
  "mcp": {
    "servers": {
      "pickfu": {
        "url": "https://mcp.pickfu.com/mcp",
        "transport": "streamable-http"
      }
    }
  }
}
Then restart the gateway. OAuth sign-in is triggered when a tool is first called. See OpenClaw MCP documentation for details.

​
CLI inside a skill

Install the PickFu CLI on the machine running OpenClaw (see Installation for all methods — Homebrew, curl, npm, or standalone binary), then reference pickfu commands from your skill like any other shell tool. The CLI is a good fit for short-lived operations that slot cleanly into OpenClaw’s execution model: creating and publishing surveys, generating or uploading media, tagging and organizing projects, and retrieving results from surveys that have already finished. pickfu survey create, pickfu media upload, pickfu media generate, and pickfu survey get all return quickly and work well inside a skill.
pickfu survey watch isn’t a good fit for OpenClaw skills. PickFu surveys typically take 15–30 minutes or longer to collect responses, and sitting on a blocking watch call that long is at or past OpenClaw’s exec tool timeout. For long blocking waits, use the PickFu CLI from Claude Code or another runtime without a hard tool timeout. In OpenClaw, prefer a fire-and-resume pattern: publish the survey in one skill invocation, exit, and call pickfu survey get <survey-id> in a follow-up once you expect responses to be in.

​
Ready-made: PickFu Market Research skill

The PickFu Market Research skill on Claw Hub is a ready-to-install workflow built on top of the PickFu CLI. Install it to get:
  • Prompts for generating logo concepts, packaging mockups, and ad creatives
  • Templates for validating product names and comparing packaging designs
  • A generate → test → refine → retest iteration loop using real consumer feedback
  • Tagging and project organization helpers for managing multiple research threads
The skill wraps the PickFu CLI, so you’ll need the CLI installed on the machine running OpenClaw. See the CLI page for install methods.

​
Need help?

OpenClaw docs

Runtime, skills, gateway setup, and MCP client configuration.

PickFu MCP server

MCP server reference, tool catalog, and authentication.

PickFu CLI

Install methods, all commands, and flag reference.

Contact support

Get help with OpenClaw + PickFu integration.