Apify Plugin for OpenClaw

Universal web scraping and data extraction via Apify — 20k+ Actors across Instagram, Facebook, TikTok, YouTube, Google Maps, Google Search, e-commerce, and more.

Install

openclaw plugins install @apify/apify-openclaw-plugin

Restart the Gateway after installation.

Upgrading from 0.1.x to 0.2.x

Version 0.2.0 renamed the plugin id from apify to apify-openclaw-plugin to match the unscoped npm package name. If you installed 0.1.x, openclaw plugins update apify will fail with:

Failed to update apify: plugin id mismatch: expected apify, got apify-openclaw-plugin

Download and run the migration script once before updating:

curl -fsSL https://raw.githubusercontent.com/apify/apify-openclaw-plugin/main/scripts/migrate-id.mjs -o /tmp/apify-migrate-id.mjs
node /tmp/apify-migrate-id.mjs

(Inspect /tmp/apify-migrate-id.mjs first if you prefer.)

Then update normally:

openclaw plugins update apify-openclaw-plugin

The script reads ~/.openclaw/openclaw.json (or $OPENCLAW_CONFIG_PATH), renames the apify id to apify-openclaw-plugin in plugins.installs, plugins.entries, plugins.allow, plugins.deny, and plugins.slots.memory, then writes the file back. It is safe to re-run.

Note: if your config is JSON5 (has comments or trailing commas), the script will refuse to parse it — apply the rename by hand instead.

How it works

The plugin registers a single tool — apify — with three actions:

The tool uses a two-phase async pattern: start fires off a run and returns immediately. collect fetches results when the run completes. The agent does other work in between.

Get an API key

1. Create an Apify account at https://console.apify.com/
2. Generate an API token in Account Settings → Integrations.
3. Store it in plugin config or set the APIFY_API_KEY environment variable.

Configure

{
  plugins: {
    entries: {
      "apify-openclaw-plugin": {
        config: {
          apiKey: "apify_api_...",     // optional if APIFY_API_KEY env var is set
          baseUrl: "https://api.apify.com",
          maxResults: 20,
          enabledTools: [],           // empty = all tools enabled
        },
      },
    },
  },
  // Make the tool available to agents:
  tools: {
    alsoAllow: ["apify"],   // or "apify" or "group:plugins"
  },
}

Or use the interactive setup wizard:

openclaw apify setup

apify

Workflow

discover (search) → discover (schema) → start → collect

1. Search — Find Actors: { action: "discover", query: "amazon price scraper" }
2. Schema — Get input params: { action: "discover", actorId: "apify~google-search-scraper" }
3. Start — Run the Actor: { action: "start", actorId: "apify~google-search-scraper", input: { queries: ["OpenAI"] } }
4. Collect — Get results: { action: "collect", runs: [{ runId: "...", actorId: "...", datasetId: "..." }] }

Actor ID format

Actor IDs use the username~actor-name format (tilde separator, not slash).

Known Actors

The tool description includes 20k+ Actors across these categories:

• Instagram — profiles, posts, comments, hashtags, reels, search, followers, tagged posts
• Facebook — pages, posts, comments, likes, reviews, groups, events, ads, reels, photos, marketplace
• TikTok — search, profiles, videos, comments, followers, hashtags, sounds, ads, trends, live
• YouTube — search, channels, comments, shorts, video-by-hashtag
• Google Maps — places, reviews, email extraction
• Other — Google Search, Google Trends, Booking.com, TripAdvisor, contact info, e-commerce

Batching

Most Actors accept arrays of URLs/queries in their input (e.g., startUrls, queries). Always batch multiple targets into a single run — one run with 5 URLs is cheaper and faster than 5 separate runs.

Examples

// 1. Search the Apify Store
const search = await apify({
  action: "discover",
  query: "linkedin company scraper",
});

// 2. Get an Actor's input schema
const schema = await apify({
  action: "discover",
  actorId: "compass~crawler-google-places",
});

// 3. Start a Google Search scrape
const started = await apify({
  action: "start",
  actorId: "apify~google-search-scraper",
  input: { queries: ["OpenAI", "Anthropic"], maxPagesPerQuery: 1 },
  label: "search",
});
// -> { runs: [{ runId, actorId, datasetId, status }] }

// 4. Collect results
const results = await apify({
  action: "collect",
  runs: started.runs,
});
// -> { completed: [...], pending: [...] }

// Instagram profile scraping
await apify({
  action: "start",
  actorId: "apify~instagram-profile-scraper",
  input: { usernames: ["natgeo", "nasa"] },
});

// TikTok search
await apify({
  action: "start",
  actorId: "clockworks~tiktok-scraper",
  input: { searchQueries: ["AI tools"], resultsPerPage: 20 },
});

Sub-agent delegation

The tool description instructs agents to delegate apify calls to a sub-agent. The sub-agent handles the full discover → start → collect workflow and returns only the relevant extracted data — not raw API responses or run metadata.

Security

• API keys are resolved from plugin config or APIFY_API_KEY env var — never logged or included in output.
• Base URL validation — only https://api.apify.com prefix is allowed (SSRF prevention).
• External content wrapping — all scraped results are wrapped with untrusted content markers.

Development

# Install dependencies
npm install

# Type check
npm run typecheck

# Run tests
npm test

# Build compiled output to dist/ (required for publish)
npm run build

# Pack (dry run) — npm runs `prepublishOnly` (build) automatically before packing
npm pack --dry-run

dist/ is generated by npm run build and is not checked in. The published npm tarball ships dist/ so newer OpenClaw versions (which no longer JIT-load TypeScript) can install the plugin.

Support

For issues with this integration, contact integrations@apify.com.

License

MIT