The PickFu Model Context Protocol (MCP) integration lets you create and manage polls directly from AI assistants like Cursor and Claude Desktop. Once connected, you can build surveys, manage targeting, retrieve results, and search the help center without leaving your AI workflow.
What is MCP?
Model Context Protocol (MCP) is an open standard that connects AI assistants to external tools and data sources. The PickFu MCP server gives your AI assistant direct access to PickFu's polling platform, enabling you to:
Create and publish polls programmatically
Retrieve survey results and status
Search available targeting options and reporting fields
Manage tags and organize polls
Search the PickFu help center for guidance
Benefits of the MCP integration
Streamlined workflow — Create polls without switching between your IDE or editor and the PickFu dashboard
Rapid iteration — Test multiple variations of questions, images, or targeting parameters in minutes
Natural language control — Describe what you want to test in plain English; your AI assistant handles the API details
Context-aware polling — Your AI assistant can suggest poll types, targeting options, and questions based on your project
Before you begin
You'll need a PickFu account (any plan) and an MCP-compatible AI assistant (Cursor or Claude Desktop).
Connect to Cursor
To add the PickFu MCP server to Cursor:
Open Cursor and go to Settings (Cmd/Ctrl + ,)
Navigate to Features > Model Context Protocol
Click Add Server and select Custom MCP Server
Enter the server URL:
https://mcp.pickfu.com/mcpClick Save
When prompted, authenticate with your PickFu account through the OAuth flow
You'll be redirected back to Cursor once authentication is complete
Cursor will verify the connection and list available PickFu tools in your MCP panel.
Connect to Claude Desktop
To configure the PickFu MCP server in Claude Desktop:
Locate your Claude configuration file:
macOS:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
Open the file in a text editor and add the PickFu server to the
mcpServerssection:
{
"mcpServers": {
"pickfu": {
"url": "https://mcp.pickfu.com/mcp"
}
}
}Save the file and restart Claude Desktop
When Claude connects to the MCP server, you'll be prompted to authenticate with your PickFu account
Complete the OAuth login in your browser, then return to Claude Desktop
Once authenticated, PickFu tools will be available in your conversations.
Available tools and capabilities
Once connected, your AI assistant can access these PickFu tools:
build_poll_url
Generates a pre-filled PickFu poll URL with your specified parameters. Useful for quickly opening polls in your browser with targeting, questions, and options already configured.
Parameters: Poll type, question text, options (text/media/ASINs), targeting criteria, sample size
create_survey
Creates a new poll in draft mode. Returns a survey ID you can use with other tools to publish or update the poll.
Parameters: Poll type, question, options, targeting attributes, respondent count
publish_survey
Publishes a draft survey and starts collecting responses. Once published, the poll enters the respondent queue and begins charging your account.
Parameters: Survey ID
get_survey
Retrieves details and results for a specific poll, including status, response count, completion percentage, and respondent feedback.
Parameters: Survey ID
update_survey
Modifies an unpublished survey. You can change the question, options, targeting, or sample size before publishing.
Parameters: Survey ID, updated fields
list_surveys
Returns a list of your recent polls with status, creation date, and basic metadata. Use this to find survey IDs or track active polls.
Parameters: Optional filters (status, date range, tags)
list_available_targeting
Shows all demographic and psychographic targeting attributes you can use when creating polls. PickFu offers 100+ targeting traits including age, location, income, interests, and shopping behaviors.
Parameters: None
list_available_reporting
Lists available reporting dimensions and metrics for analyzing poll results, such as demographic breakdowns and response patterns.
Parameters: None
list_tags
Retrieves all tags you've created for organizing polls. Tags help you categorize surveys by project, client, or test type.
Parameters: None
create_tag
Creates a new tag for organizing your polls.
Parameters: Tag name
help_center_search
Searches the PickFu help center for articles matching your query. Useful when you need guidance on poll best practices, targeting options, or interpreting results.
Parameters: Search query
Supported poll types
The MCP integration supports all PickFu poll formats:
head-to-head — A/B comparison of two options (most common for quick decisions)
ranked — Rank 3-8 options from most to least preferred
open-ended — Free-text responses to a question (qualitative research)
click-test — Heatmap showing where respondents click on an image
image-set — Collect reactions to multiple images (emojis or star ratings)
single-select — Choose one option from a list
multi-select — Choose multiple options from a list
amazon-mockup — Test Amazon listing variations with realistic product page previews
Option formats
When creating polls, you can provide options in these formats:
Text — Plain text strings for questions like "Which tagline is more compelling?" or multiple-choice options
Image URLs — Direct links to images (JPG, PNG, GIF, WebP). Images are displayed at optimal size for comparison
Video URLs — Links to videos hosted on YouTube, Vimeo, or direct MP4 URLs (for video ads or app demos)
Amazon ASINs — Product identifiers (e.g.,
B08N5WRWNW) to generate realistic Amazon listing mockups. The MCP server automatically fetches product details and creates comparison views
For best results with image tests, use high-resolution images (at least 800px wide) and ensure URLs are publicly accessible. PickFu optimizes images automatically for display.
Usage examples
Example 1: A/B test two product images
Ask your AI assistant:
"Create a head-to-head poll comparing these two product images for women aged 25-45 in the US who shop on Amazon. Use 50 respondents. Image A: https://example.com/product-a.jpg and Image B: https://example.com/product-b.jpg. Question: Which product image makes you more likely to purchase?"
The assistant will use create_survey with poll type head-to-head, your image URLs as options, and targeting for gender: female, age: 25-45, location: United States, shopping_behavior: amazon_shopper. It then publishes the poll and returns the survey ID.
Example 2: Compare Amazon listings by ASIN
Ask your AI assistant:
"I want to test my Amazon listing against two competitors. My ASIN is B08N5WRWNW, competitors are B07XJ8C8F5 and B09GF3QZJX. Create a ranked poll for 100 Amazon shoppers interested in home organization."
The assistant uses create_survey with poll type ranked, the three ASINs as options, targeting for shopping_behavior: amazon_shopper, interests: home_organization, and publishes for 100 respondents. PickFu generates realistic Amazon mockups automatically.
Example 3: Iterate on poll targeting
Ask your AI assistant:
"Show me all available targeting options related to gaming and create a click test poll testing this UI screenshot with male gamers aged 18-35."
The assistant calls list_available_targeting, filters for gaming-related attributes (e.g., interests: video_games, gaming_platform: pc, console_ownership), then uses create_survey with poll type click-test, your screenshot URL, and the refined targeting criteria.
Troubleshooting
Connection failed or authentication error
Symptom: Your AI assistant reports "authentication required" or "connection refused" when trying to use PickFu tools.
Solution:
Ensure you completed the OAuth authentication flow when first connecting
Check that the MCP server URL is exactly
https://mcp.pickfu.com/mcpwith no trailing slashRestart your AI assistant to trigger a new authentication attempt
Clear your browser cookies for pickfu.com and re-authenticate
Session expired or token invalid
Symptom: Tools worked previously but now return "invalid token" or "session expired" errors.
Solution:
OAuth sessions can expire after extended periods of inactivity. Restart your AI assistant to trigger re-authentication
If prompted, log in to PickFu again through the OAuth flow
Check that you're still logged in to your PickFu account in your browser
Rate limit exceeded
Symptom: Your AI assistant reports "rate limit exceeded" or "too many requests."
Solution:
The MCP server limits requests to prevent abuse. Wait 60 seconds before retrying
Batch operations when possible (e.g., use
list_surveysonce instead of callingget_surveyrepeatedly)Contact support if you need higher rate limits for enterprise use
Survey creation failed
Symptom: create_survey returns an error or validation message.
Solution:
Invalid poll type: Check that you're using a supported type (head-to-head, ranked, open-ended, etc.) with the exact identifier shown in the "Supported poll types" section
Invalid options: Ensure image/video URLs are publicly accessible and ASINs are valid 10-character Amazon identifiers
Targeting mismatch: Use
list_available_targetingto verify attribute names and valid values. Targeting criteria must match PickFu's demographic options exactlySample size out of range: PickFu supports 15-500 respondents per poll. Adjust your requested sample size to fall within this range
Tool not found or unavailable
Symptom: Your AI assistant says "PickFu tool not available" or doesn't recognize MCP commands.
Solution:
Verify the MCP server is properly configured in your AI assistant's settings
Check that the server URL is exactly
https://mcp.pickfu.com/mcpwith no trailing slashRestart your AI assistant to reload MCP connections
Confirm your version of Cursor or Claude Desktop supports MCP (update if necessary)
Insufficient credits or account issue
Symptom: publish_survey fails with "insufficient credits" or "account suspended."
Solution:
Check your account balance in the PickFu dashboard. Add credits or update your payment method if needed
Ensure your account is in good standing. Contact support if you see suspension notices
Verify you're using
create_survey(draft mode, no charge) beforepublish_survey(charges account) to preview costs
Frequently asked questions
Do I need a paid PickFu plan to use the MCP integration?
No. The MCP integration is available to all PickFu accounts, including free accounts. You'll still need credits to publish polls, but you can connect and use the MCP tools with any account type.
Can I use the MCP integration with other AI assistants besides Cursor and Claude Desktop?
Any MCP-compatible AI assistant or tool can connect to the PickFu MCP server. The setup process may differ slightly from the examples above, but the server URL and authentication method remain the same.
Are there usage limits or rate limits for the MCP integration?
Yes. The MCP server enforces rate limits to ensure fair usage. You can create up to 100 API requests per minute. If you exceed this limit, wait 60 seconds before retrying. Enterprise accounts can request higher limits by contacting support.
How do I know which targeting options are available for my poll?
Use the list_available_targeting tool to see all 100+ demographic and psychographic attributes PickFu supports. Your AI assistant can filter and suggest relevant options based on your poll's purpose.
Can I retrieve results for polls created in the PickFu dashboard?
Yes. The list_surveys and get_survey tools work with any poll in your account, regardless of how it was created. Use list_surveys to find the survey ID, then get_survey to retrieve results.
What happens if I publish a poll and want to cancel it?
Once published, polls cannot be canceled through the MCP integration. Log in to your PickFu dashboard and contact support to stop a running poll. Note that you're charged for completed responses even if you cancel early.