AI Agent Integration — MCP install, Actor-as-tool, AI scaffolding, schema gen, diagnose

[kuntal1461]

Overview

Make Apify the default data and automation layer for AI agents and AI IDEs. This Epic covers five features that together lower the barrier to connecting Apify Actors with AI workflows — from one-command MCP setup, to AI-generated Actor scaffolds, to cloud log diagnosis.

Why this matters

• MCP is the dominant protocol for AI agent tooling (Claude, Cursor, Windsurf, Copilot all support it)
• Apify already has apify-mcp-server and thousands of Actors — the infrastructure is there
• The missing piece is CLI ergonomics: making it trivial for developers to connect, build, publish, and debug Actors as AI-native tools
• Each Actor run triggered by an AI agent = direct compute revenue

---

Sub-tasks

• 1. feat: apify mcp install — one-command MCP setup for Claude, Cursor, Windsurf, and Copilot #1438 apify mcp install — one-command MCP setup for Claude, Cursor, Windsurf, Copilot
• 2. feat: expose any Actor as an MCP tool — apify actors expose-as-mcp #1439 apify actors expose-as-mcp — expose any Actor as an MCP tool with one flag
• 3. feat: apify create --from-url <website> — AI generates Actor scaffold from a target URL #1440 apify create --from-url <url> — AI generates Actor scaffold from a target website
• 4. feat: apify actor generate-schema — AI generates INPUT_SCHEMA.json from existing Actor code #1441 apify actor generate-schema — AI generates INPUT_SCHEMA.json from existing code
• 5. feat: apify diagnose — AI reads cloud run error logs and suggests fixes #1442 apify diagnose — AI reads cloud run error logs and suggests fixes

---

Recommended build order

1. feat: apify mcp install — one-command MCP setup for Claude, Cursor, Windsurf, and Copilot #1438 (apify mcp install) — no LLM needed, pure config-file UX, unblocks everything else
2. feat: expose any Actor as an MCP tool — apify actors expose-as-mcp #1439 (expose-as-mcp) — natural next step once MCP install exists
3. feat: apify actor generate-schema — AI generates INPUT_SCHEMA.json from existing Actor code #1441 (generate-schema) — reduces publish friction, feeds into feat: expose any Actor as an MCP tool — apify actors expose-as-mcp #1439 tool descriptions
4. feat: apify diagnose — AI reads cloud run error logs and suggests fixes #1442 (diagnose) — needs LLM API decision, builds on feat: Add apify actors doctor for local Actor project diagnostics #1366 (actors doctor)
5. feat: apify create --from-url <website> — AI generates Actor scaffold from a target URL #1440 (create --from-url) — highest impact but most complex; tackle last

---

Dependencies / Related

• apify/apify-mcp-server — MCP server that feat: apify mcp install — one-command MCP setup for Claude, Cursor, Windsurf, and Copilot #1438 and feat: expose any Actor as an MCP tool — apify actors expose-as-mcp #1439 configure
• apify/mcpc — universal MCP CLI client
• feat: Add apify actors doctor for local Actor project diagnostics #1366 — apify actors doctor (offline pre-flight checks, ships before feat: apify diagnose — AI reads cloud run error logs and suggests fixes #1442)
• Guided Actor creation flow in apify create (parity with Console wizard) #1153 — Guided apify create wizard (UX foundation for feat: apify create --from-url <website> — AI generates Actor scaffold from a target URL #1440)
• Multi-token / multi-account handling in the CLI #1297 — Multi-account support (needed for multi-IDE MCP configs in feat: apify mcp install — one-command MCP setup for Claude, Cursor, Windsurf, and Copilot #1438)

---

Open questions (for PM review)

• Which LLM / API for items 3, 4, 5 — Apify-hosted model or user brings their own key?
• Should AI features be gated behind a flag (--ai) or on by default?
• Priority order — does the team agree with the build order above?

[kuntal1461]

Hi @patrikbraborec — would love to get your take on this whenever you have a moment. There are a few open questions at the bottom that really need a product perspective to move forward — particularly around the LLM approach and whether AI features should be opt-in or on by default. Curious whether the overall direction feels aligned with where the team is headed, and happy to reprioritize or reshape any of the sub-tasks based on what makes the most sense for Apify!