Fodda MCP Server

Expert-curated knowledge graphs for AI agents — PSFK Retail, Beauty, Sports and partner datasets via the Model Context Protocol.

Quick Start

Claude (Web — Pro, Max, Team, Enterprise)

1. Get your personal MCP URL at app.fodda.ai → Account → MCP Integration (format: https://mcp.fodda.ai/c/<your-token>)

2. In Claude, go to Settings → Connectors → Add custom connector, paste the URL, and click Add

3. Start chatting with your Fodda knowledge graphs

Note: legacy ?api_key= URLs are no longer accepted (raw keys in URLs leak into logs and browser history). If your connector stops working, just grab a fresh MCP URL at app.fodda.ai. One-click OAuth connect arrives with Fodda's Anthropic Connectors Directory listing.

Claude Code (CLI — SSE)

claude mcp add --transport sse fodda https://mcp.fodda.ai/sse \
  --header "Authorization: Bearer YOUR_API_KEY"

Claude Enterprise

For Claude Enterprise with admin-managed connectors, your workspace admin can register the Fodda MCP server using the same Streamable HTTP endpoint (https://mcp.fodda.ai/mcp) via the Admin Console. See Enterprise MCP Setup for full details.

OpenAI (Responses API — Recommended)

Use OpenAI's native MCP support — no function definitions needed:

from openai import OpenAI
client = OpenAI()

resp = client.responses.create(
    model="gpt-5",
    tools=[{
        "type": "mcp",
        "server_label": "fodda",
        "server_description": "Expert-curated knowledge graphs for retail, beauty, sports trends and earnings intelligence.",
        "server_url": "https://mcp.fodda.ai/mcp",
        "authorization": "Bearer YOUR_API_KEY",
        "require_approval": "never",
    }],
    input="What are the top emerging retail trends?",
)
print(resp.output_text)

import OpenAI from "openai";
const client = new OpenAI();

const resp = await client.responses.create({
  model: "gpt-5",
  tools: [{
    type: "mcp",
    server_label: "fodda",
    server_description: "Expert-curated knowledge graphs for retail, beauty, sports trends and earnings intelligence.",
    server_url: "https://mcp.fodda.ai/mcp",
    authorization: "Bearer YOUR_API_KEY",
    require_approval: "never",
  }],
  input: "What are the top emerging retail trends?",
});
console.log(resp.output_text);

Tip: To reduce latency with 30+ tools, use allowed_tools to filter: "allowed_tools": ["search_graph", "search_statistics", "brand_tracker"]

Legacy: If you're using chat.completions.create() with function calling, define a custom function that calls the Fodda REST API. See fodda.ai/connect for details.

Generic Streamable HTTP Client

Connect to the /mcp endpoint using HTTP POST (Streamable HTTP) with Authorization: Bearer YOUR_API_KEY:

https://mcp.fodda.ai/mcp

Gemini CLI

{
  "tools": [{
    "type": "mcp",
    "name": "fodda",
    "url": "https://mcp.fodda.ai/sse",
    "headers": { "Authorization": "Bearer YOUR_API_KEY" }
  }]
}

Generic SSE Client

Connect to https://mcp.fodda.ai/sse with an Authorization: Bearer YOUR_API_KEY header.

Related MCP server: AgentDB

Available Tools

The Fodda MCP server exposes a rich suite of 30 tools to search, analyze, brainstorm, and visualize trends, expert-curated knowledge graphs, and corporate documents.

Discovery & Search

Traversal & Graph Operations

Intelligence Domains

Earnings & Corporate Insights

Brand & Ideation

Deep Research & Execution Agents

Synthetic Analyst & Visualization

Note on Dynamic Partner Skills: When partner skills (such as Paralogy or Igloo) are enabled for a user's account, additional specialized tools (e.g. paralogy_analyze_trends, ideation tools) are dynamically registered on the MCP server and made available automatically.

Note on System Utilities: The server also registers several helper and account management utilities (e.g., get_my_account, toggle_graph_preference, update_user_profile, manage_scheduled_reports, send_feedback) to allow the AI assistant to query subscription status, manage scheduled briefings, or save personalization settings directly.

Discovery Endpoints

Authentication

Preferred — HTTP header (used by the directory connection and all header-capable clients). Pass your Fodda API key as a Bearer token:

Authorization: Bearer sk_live_...

The server also accepts the key via an X-API-Key: sk_live_... header.

In MCP request _meta:

{ "_meta": { "authorization": "Bearer sk_live_..." } }

Fallback — URL query parameter. Some clients (e.g. the Claude.ai web custom connector UI) cannot set custom request headers. For those, the key may be passed as ?api_key=.... This is provided only for compatibility; prefer header-based auth wherever the client supports it, since URLs are more likely to be logged by intermediaries. The server never returns your API key in tool output.

Configuration

Build & Run

npm install
npm run build

# Stdio mode
npm start

# SSE mode
PORT=8080 npm start

Self-Hosting

• Docker: docker build -t fodda-mcp . && docker run -p 8080:8080 -e PORT=8080 fodda-mcp

• Cloud Run: ./deploy_cloud_run.sh

• Kubernetes: See deployment/k8s/

• Terraform: See deployment/terraform/

MCP Registry

This server is published to the Official MCP Registry as ai.fodda/mcp-server.

# Verify listing
curl "https://registry.modelcontextprotocol.io/v0.1/servers?search=ai.fodda/mcp-server"

Privacy Policy

Full policy: https://www.fodda.ai/privacy

Summary of how the Fodda MCP server handles data (the hosted policy is authoritative):

• What we collect. Account information (name, email, organization) to manage your account; usage data — API queries (query text, graph accessed, timestamps) logged for billing, service improvement, and abuse prevention; and standard technical data (IP, browser type, referral URLs). We do not store the full content of AI-generated responses. Your API key authenticates requests and is never returned in tool output.

• How it's used. To provide and improve the Services, process billing, enforce rate limits and prevent abuse, communicate support/service updates, and meet legal obligations. Requests are proxied to the Fodda API (https://api.fodda.ai) over TLS.

• Third-party sharing. We do not sell your personal information. Data is shared only with service providers (e.g. payment processors, cloud hosting) under confidentiality agreements, or when required by law.

• AI model training. Fodda does not use your queries or data to train AI models. Knowledge graphs are expert-curated, not generated from user interactions.

• Retention. Account information is retained while your account is active; query logs are retained for billing and analytics. You may request account/data deletion via the contact below.

• Security. TLS-encrypted connections, API-key authentication, and secure cloud infrastructure.

• Contact. Privacy inquiries: privacy@fodda.ai.

See the full hosted Privacy Policy (last updated June 16, 2026) for the complete, authoritative terms.

Support

• Email: hello@fodda.ai

• Account & API keys: app.fodda.ai → Account → MCP Integration

• Documentation: fodda.ai

To report a security issue, email hello@fodda.ai with "SECURITY" in the subject line.

License

Proprietary — fodda.ai