Model Context Protocol (MCP)

Connect AI Assistants to Your Affise Platform

Query statistics, search offers, analyze conversions and partners using natural language. Connect to our hosted remote server in seconds, or run it locally — your choice.

⚡ Connect Remote Server See How It Works

🌐 Hosted remote endpoint 🔑 OAuth 2.0 & Bearer 💻 Local stdio option

ask your AI assistant

// Ask your AI assistant:
> "Show me the top 10 offers
  by revenue this week"

{
  "tool": "affise_stats",
  "params": {
    "date_from": "2026-05-26",
    "date_to":   "2026-06-01",
    "order":    "income",
    "limit":    10
  }
}

Two ways to connect

Remote or Local — Same Tools, Your Choice

Connect to the Affise-hosted server over HTTP with OAuth or a bearer token, or run the open-source server locally over stdio. Both expose the exact same MCP toolset.

Recommended · Hosted

🌐 Remote MCP Server

A managed, always-on HTTP endpoint. No install, no local runtime — just point your MCP client at the URL and authenticate.

→Endpoint: https://mcp.affise.com/mcp
→OAuth 2.0 browser sign-in (per-user, encrypted session)
→or a static Bearer token for CI & scripts
→Works in Claude.ai, Claude Desktop, ChatGPT, Cursor & more
→Rate-limited, session-managed, always up to date

Connect the hosted server

Self-hosted

💻 Local stdio Server

Run the open-source server on your own machine for full control. Communicates with the Affise API directly from your environment.

→Install via npm: @affise/mcp-server (stdio)
→Auth with your Affise API key via env vars
→Ideal for Claude Desktop, Cursor & Claude Code
→Data flows only between your host and Affise
→Fully auditable — MIT-licensed source

Get it on npm

Compatible With

Works with Your Favorite AI Tools

The Affise MCP Server implements the open Model Context Protocol standard — compatible with any MCP-enabled client, remote or local.

◎

Claude.ai & Desktop

Add the remote URL or run locally — native MCP support for seamless data querying.

◆

Cursor IDE

AI-powered code editor that can reach your Affise data while you build.

◉

Claude Code & ChatGPT

CLI agent and ChatGPT connectors — both attach to the hosted remote endpoint.

◈

Any MCP Client

Open standard — connect with Windsurf, Cline, Continue and more.

Features

Everything You Need to Query Affise with AI

Access your affiliate-marketing data through natural-language conversations — across 23 purpose-built MCP tools.

📊

Statistics & Reports

Pull real-time stats sliced by date, offer, affiliate, advertiser, country, sub-IDs and more. Reports in one prompt.

🎯

Offer Discovery

List, search and inspect offers — caps, payouts, targeting, categories and tracking links — conversationally.

👥

Affiliate & Advertiser Data

Look up partner and advertiser details, performance history and approval status. Compare side by side.

🔄

Conversion Tracking

Query raw and aggregated conversions, analyze trends and trafficback, and debug tracking with AI.

📈

Retention & TTA Analytics

Ask about retention rates and time-to-action — get instant, data-backed answers from your instance.

🤝

Partner Self-Service

Partner-scoped tools: profile, balance, news, visible offers, live offers and sub-ID discovery.

How It Works

Up and Running in 3 Minutes

Pick remote or local, authenticate, and start asking questions.

Add the Server

Point your MCP client at the hosted URL https://mcp.affise.com/mcp, or install @affise/mcp-server from npm to run it locally.

Authenticate

Remote: sign in with OAuth 2.0 or paste a Bearer token. Local: set your Affise API key and base URL.

Start Asking Questions

Open your AI assistant and ask anything about your Affise data — stats, offers, partners, conversions.

23

MCP Tools Available

2

Transports — Remote & Local

<200ms

Average Response Time

OSS

Open Source & Free

Security & Privacy

Enterprise-Grade Security Built In

Choose where the server runs. The hosted endpoint adds OAuth, encrypted per-user sessions and rate limiting; the local server keeps everything inside your own environment.

🔑

OAuth 2.0 & Bearer Auth

The remote server supports an OAuth browser flow with encrypted per-user sessions, plus static bearer tokens for automation.

🏠

Local Option, Zero Relay

Prefer no third party in the path? Run the open-source server locally — data flows only between your host and the Affise API.

🛡

Read-Only Surface

The public toolset is read-only by design — safe for analytics and reporting without risk of accidental changes.

✅

SOC 2 Type II Platform

The Affise platform is SOC 2 certified; the MCP server inherits the same posture via scoped API access.

claude_desktop_config.json — remote

{
  "mcpServers": {
    "affise": {
      "type": "http",
      "url": "https://mcp.affise.com/mcp",
      "headers": {
        "Authorization": "Bearer <token>"
      }
    }
  }
}

# or just add the URL in Claude.ai
# → Settings ▸ Connectors ▸ Add MCP
# → sign in via OAuth, done.

Get Started

Connect in Seconds

Use the hosted remote endpoint for the fastest path, or self-host from GitHub. Local server runs on Node.js 18+.

Open Remote Endpoint Talk to an Expert

Terminal — local (self-hosted)

# Install via npm (or run with npx)
$ npm install -g @affise/mcp-server

# Set your API credentials
$ export AFFISE_API_KEY="your-api-key"
$ export AFFISE_BASE_URL="https://api.affise.com"

# Run the server (stdio)
$ affise-mcp-server
✔ MCP server listening on stdio
✔ Connected to Affise API
✔ 23 tools registered