SEO Analyzer

What it does

Analyzes the on-page SEO of public web pages. For each URL it fetches the page with a plain HTTP GET (no JavaScript rendering, no headless browser), parses the HTML, and delivers one record with the title, meta description, canonical link, robots meta, headings, Open Graph tags, language, word count, and a list of detected issues from a documented rule set.

Inputs

Field	Type	Required	Default	Description
urls	array of strings	yes	["https://example.com"]	Public page URLs to analyze. Maximum 500 per run.

Outputs

One dataset record per URL. status: "ok" records carry the full analysis; status: "error" records (transport failures: DNS, TCP, TLS, timeout, invalid URL) carry only url, errorMessage, fetchedAt.

Field	Type	Description
status	string	ok (analysis delivered) or error (fetch failed).
url	string	The requested URL.
finalUrl	string	Final URL after redirects.
statusCode	integer	HTTP status of the final response.
responseTimeMs	integer	Fetch wall time in milliseconds.
title	string | null	<title> contents.
metaDescription	string | null	meta[name=description].
canonical	string | null	link[rel=canonical] href.
robotsMeta	string | null	meta[name=robots].
h1	array	Text of every <h1>.
h2Count	integer	Number of <h2> elements.
ogTitle / ogDescription / ogImage	string | null	Open Graph tags.
lang	string | null	lang attribute of <html>.
wordCount	integer	Visible body word count (script/style removed).
issues	array	Issue codes (see rule set below).
errorMessage	string	[error] Why the fetch failed.
fetchedAt	string	ISO 8601 timestamp.

Issue rule set: missing-title, title-too-short (<30 chars), title-too-long (>60 chars), missing-meta-description, missing-h1, multiple-h1, missing-canonical, non-200-status.

Example record:

{
    "status": "ok",
    "url": "https://example.com",
    "finalUrl": "https://example.com/",
    "statusCode": 200,
    "responseTimeMs": 83,
    "title": "Example Domain",
    "metaDescription": null,
    "canonical": null,
    "robotsMeta": null,
    "h1": ["Example Domain"],
    "h2Count": 0,
    "ogTitle": null,
    "ogDescription": null,
    "ogImage": null,
    "lang": null,
    "wordCount": 24,
    "issues": ["title-too-short", "missing-meta-description", "missing-canonical"],
    "fetchedAt": "2026-10-12T08:46:13.717Z"
}

Pricing

Pay-per-event: one result event per analyzed URL — i.e. per record with status: "ok". Records with status: "error" (failed fetches) are delivered free. You are never charged for a URL we could not fetch, and a run that delivers zero analyses fails loudly instead of charging anything.

Final per-result price is set at publish under the Gate 1 rule (>= 3x measured platform cost, <= usage-dominant incumbent's per-item price).

Limits

• Plain HTTP GET only — no JavaScript rendering, no headless browser. Pages that require JS to render will be analyzed as their raw HTML.
• Maximum 500 URLs per run; 20s per-request timeout; up to 2 retries with backoff on network errors and HTTP 5xx (4xx is not retried).
• A persistent 5xx is still an HTTP response: it is delivered as an ok record carrying the non-200-status issue.
• URLs are fetched sequentially; runtime grows roughly linearly with URL count.
• If every URL fails to fetch, the run fails with a clear status message.

Changelog

0.0

• Initial release.

---

House conventions (not actor.json fields — set on the platform at publish): pay-per-event monetization only; no Standby mode; limited permissions only. Charge via Actor.charge({ eventName }) ONLY after a result is delivered; never charge on failures. Fail loudly (Actor.fail) when zero results are delivered, so broken runs never look successful.