About Wishfinity +w
Wishfinity +w is an MCP server for saving product links to a user's Wishfinity wishlist from AI shopping and gift conversations. It returns a link the user clicks to save an item, requires no API key, and supports local stdio and remote HTTP transports.
Key features
- 01Add product links to a wishlist
- 02Return a save link for user action
- 03Support stdio and HTTP transports
- 04Provide prompts and resources
Use cases
- 01Save recommended products for later purchase
- 02Collect gift candidates in a wishlist
- 03Bookmark items when postponing a purchase
- 04Integrate wishlist actions into AI applications
MCP support
The MCP server provides the add_to_wishlist tool, which accepts a product URL and returns a Wishfinity link that the user clicks to save the item. It also provides MCP prompts and resources, with stdio and HTTP transports.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
1Saved tool list · updated 3w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| url | string | Yes | The product page URL to save to the wishlist |
Details
From the project README+W MCP Server (Wishfinity)
+W is a universal "save for later" action for commerce. This MCP server lets AI assistants save any product URL to a user's Wishfinity wishlist with one click.
Works with Claude, ChatGPT, Gemini, LangChain, OpenAI Agents SDK, and any MCP-compatible client.
What it does
When an AI recommends a product, it can offer +W Add to Wishlist. The user clicks the link, and the product is saved to their Wishfinity account — ready for later purchase or gifting.
User: "Find me a good espresso machine under $200"
AI: Here are 3 options...
[+W Add to Wishlist] [View on Amazon]
Zero-dependency option: /add-wish
Don't need the full MCP server? The /add-wish skill teaches any AI agent the +W pattern using just a URL — no npm, no server, no setup:
https://wishfinity.com/add?url={any_product_url}
→ View the /add-wish skill file
Works with any AI platform. The MCP server below adds richer tool integration, but /add-wish is all you need to get started.
Quick start
Option 1: Local installation (stdio transport)
Best for Claude Desktop, ChatGPT Desktop, Cursor, VS Code, and local development.
npm install wishfinity-mcp-plusw
Add to your MCP client configuration:
{
"mcpServers": {
"wishfinity": {
"command": "npx",
"args": ["wishfinity-mcp-plusw"]
}
}
}
Option 2: Remote endpoint (HTTP transport)
Best for server-side agents, LangChain production deployments, and hosted AI applications.
https://wishfinity-mcp-plusw.wishfinity.workers.dev/mcp
Platform Setup Guides
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%/Claude/claude_desktop_config.json (Windows):
{
"mcpServers": {
"wishfinity": {
"command": "npx",
"args": ["wishfinity-mcp-plusw"]
}
}
}
ChatGPT Desktop
When MCP support is available, add to your ChatGPT MCP configuration:
{
"mcpServers": {
"wishfinity": {
"command": "npx",
"args": ["wishfinity-mcp-plusw"]
}
}
}
Cursor
Add to .cursor/mcp.json in your project:
{
"mcpServers": {
"wishfinity": {
"command": "npx",
"args": ["wishfinity-mcp-plusw"]
}
}
}
LangChain
from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent
async def main():
client = MultiServerMCPClient({
"wishfinity": {
"command": "npx",
"args": ["wishfinity-mcp-plusw"],
"transport": "stdio",
}
})
tools = await client.get_tools()
agent = create_agent("openai:gpt-4", tools)
result = await agent.ainvoke({
"messages": [{"role": "user", "content": "Find me a coffee maker and save it to my wishlist"}]
})
For production (HTTP transport):
client = MultiServerMCPClient({
"wishfinity": {
"url": "https://wishfinity-mcp-plusw.wishfinity.workers.dev/mcp",
"transport": "streamable_http",
}
})
OpenAI Agents SDK
from agents import Agent, Runner
from agents.mcp import MCPServerStdio
async def main():
async with MCPServerStdio(
name="wishfinity",
params={
"command": "npx",
"args": ["wishfinity-mcp-plusw"],
},
) as server:
agent = Agent(
name="Shopping Assistant",
instructions="Help users find products and save them to wishlists.",
mcp_servers=[server],
)
result = await Runner.run(agent, "Find a good gift for a coffee lover and save it")
print(result.final_output)
Hugging Face Agents
from huggingface_hub import Agent
agent = Agent(
model="meta-llama/Llama-3.1-70B-Instruct",
mcp_servers=[{
"command": "npx",
"args": ["wishfinity-mcp-plusw"]
}]
)
response = agent.run("Find me running shoes under $150 and save to wishlist")
How it works
- AI calls the
add_to_wishlisttool with a product URL - MCP server returns a Wishfinity link:
https://wishfinity.com/add?url=... - User clicks the link → logs in (first time only) → product is saved
- User can organize, share, or purchase later
No API keys required. Authentication happens on Wishfinity's website when the user clicks the link.
Tool: add_to_wishlist
Input
| Parameter | Type | Required | Description |
|---|---|---|---|
url | string | Yes | Product page URL |
Output
| Field | Type | Description |
|---|---|---|
action_url | string | Wishfinity link to open in browser |
requires_user_action | boolean | Always true (user must click) |
display_text | string | Suggested button label |
Example
// Input
{ "url": "https://amazon.com/dp/B0EXAMPLE" }
// Output
{
"action_url": "https://wishfinity.com/add?url=https%3A%2F%2Famazon.com%2Fdp%2FB0EXAMPLE",
"requires_user_action": true,
"display_text": "Open to add to Wishfinity"
}
When to offer +W
AI clients should offer +W when:
- Explicit requests: "save for later," "add to wishlist," "bookmark this"
- After recommending products: Proactively offer to save shown items
- Indecision moments: "maybe later," "too expensive right now," "need to think about it"
- Gift context: "for my mom," "birthday gift," "perfect for my friend"
See aliases.json for the full list of trigger phrases.
Transports
| Transport | Use Case | Endpoint |
|---|---|---|
| stdio | Local clients (Claude Desktop, Cursor, etc.) | npx wishfinity-mcp-plusw |
| HTTP | Remote/server-side agents | https://wishfinity-mcp-plusw.wishfinity.workers.dev/mcp |
Button kit
The /button-kit folder contains optional UI assets (SVG icon, HTML/CSS snippets) if you want a consistent +W button appearance.
CDN URLs:
- Small:
https://cdn.jsdelivr.net/npm/wishfinity-mcp-plusw@latest/button-kit/Wishfinity-Button-Small.svg - Medium:
https://cdn.jsdelivr.net/npm/wishfinity-mcp-plusw@latest/button-kit/Wishfinity-Button-Medium.svg - Large:
https://cdn.jsdelivr.net/npm/wishfinity-mcp-plusw@latest/button-kit/Wishfinity-Button-Large.svg
Documentation
- SPEC.md — Full technical specification
- INTEGRATION_GUIDE.md — How to integrate +W into your UI
- CLOUDFLARE_SETUP.md — Deploy your own HTTP endpoint
- aliases.json — Machine-readable trigger phrases
Links
Changelog
v1.2.2 (December 24, 2025)
Critical Fix: npx execution for all developers
- Fixed main module detection to work with npx symlinks
- Resolves crash when running
npx wishfinity-mcp-plusw - Package now works flawlessly for all npm installations
v1.2.1 (December 24, 2025)
Critical Fix: MCP SDK compatibility
- Updated
@modelcontextprotocol/sdkdependency to^1.25.0 - Resolves server disconnection with SDK 1.25.1+
- Compatible with latest MCP SDK versions
v1.2.0 (December 23, 2025)
- Added MCP prompts:
save_for_later,shopping_assistant,gift_ideas - Added MCP resources:
wishfinity://guide,wishfinity://triggers - Enhanced integration capabilities for AI assistants
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.