Problem

Apify already has a world-class MCP server (apify/apify-mcp-server) and thousands of Actors. But connecting it to an AI IDE today requires a developer to:

1. Find the MCP server docs
2. Manually edit ~/.claude/settings.json, or Cursor's mcp.json, or Windsurf's config file
3. Get the right server URL / transport flags right

This is 5–10 minutes of friction that blocks non-expert users from ever connecting their AI agent to Apify.

Proposal

Add a new command:

apify mcp install [--ide claude|cursor|windsurf|copilot]

The command:

• Auto-detects which AI IDEs are installed on the machine
• Writes the correct MCP server config block into each IDE's config file
• Authenticates using the current apify login token
• Prints a confirmation with a link to test the connection

Example output

$ apify mcp install

Detected AI IDEs:
  ✓ Claude Desktop   → ~/.claude/settings.json
  ✓ Cursor           → ~/.cursor/mcp.json
  ✗ Windsurf         not installed

Installing Apify MCP server...
  ✓ Claude Desktop   updated
  ✓ Cursor           updated

Restart your IDE and ask: "Search for cat photos using Apify"
Docs: https://docs.apify.com/cli/mcp

Per-IDE config targets

Why it matters

• Removes the single biggest barrier to Apify MCP adoption: manual JSON editing
• Every Actor in the Apify Store becomes instantly accessible to AI agents
• Positions Apify as the default data/automation layer in AI IDEs
• Direct revenue channel: Agent → MCP → Actor → compute billing

Out of scope

• Modifying IDE internals or extensions
• Managing multiple Apify accounts (follow-on: multi-account support)

Related

• apify/apify-mcp-server — the MCP server this command configures
• apify/mcpc — universal MCP CLI client
• Multi-token / multi-account handling in the CLI #1297 — multi-token / multi-account handling