Scott Helvick

Apify Actors for AI agents — design notes and reference docs.

I build Apify Actors designed to be called by AI agents: callable via MCP, billable per call through x402 USDC on Base or Skyfire managed tokens, with input/output schemas tight enough that an LLM can construct correct calls from the tool description alone.

This site collects the design notes — the reasoning behind the tools, what they refuse, the tradeoffs that aren’t visible from a Store page.

Tools

See the full list at /tools/, or jump directly to:

• Stealth Browser Agent — Send a URL and a task in plain language. An LLM-driven stealth browser clicks, types, navigates, and extracts structured JSON from bot-defended sites.
• Smart Page Fetcher — Fetch a batch of URLs adaptively. Picks the cheapest method that works per URL (plain HTTP, JS render, or stealth + residential proxy). You pay per realized tier.
• Structured Extractor — Send a list of URLs and one JSON Schema. A single stealth fetch pass plus per-page LLM extraction returns schema-validated JSON per URL, with the fetch setup amortized across the batch.
• Opportunity Scout — Scores an Actor-build idea against demand and supply across nine sources. Returns PURSUE / CONSIDER / SKIP with rationale.
• Discoverability Audit — Grades any Apify Actor against an evidence-grounded rubric for AI-agent discoverability. Returns a punch list with copy-paste fixes.

Where else to find me

• Apify Store: apify.com/shelvick — live list of published Actors with current pricing, stats, and the input forms you’d use to try each one.