Website Technology Detector & Tech Stack Changes

Check supplied websites for 16 supported technologies, with evidence, CRM IDs and optional private repeat-run history. A focused HTTP-only website technology checker for bulk enrichment and agency checks—not a full Wappalyzer or BuiltWith database replacement.

Quick answer

Supply 1–50 authorized website URLs. Receive technology names, categories, the observed HTML/header signals, timestamps and an explicit status for every input. No external API key, subscription, AI service or proxy setup is required. Your normal Apify account is required.

Example

{"websites":[{"url":"https://example.com/","externalId":"customer-001"}],"authorized":true,"monitor":true,"monitorKey":"customer-checks"}

Replace the URL with an owned or authorized page. The first monitored success establishes BASELINE. Reuse the same monitor key for UNCHANGED, CHANGED or MISSING_PENDING observations. Snapshot mode is the default and does not retain named history.

Supported technologies

WordPress, Drupal, Ghost, Hugo, Joomla, Next.js, Webflow, Shopify, Google Tag Manager, Google Analytics, Stripe, Hotjar, Intercom, Cloudflare, nginx and Apache. Independently authored signals use explicit generator metadata, specific script/stylesheet URLs, framework payloads or response headers. Coverage is intentionally narrow. An absent signal never proves that a technology is absent from the entire site. Marketing text and vendor names in ordinary hyperlinks do not count as installed technology.

Outputs and CRM workflow

One dataset row per input: externalId, inputIndex, requested/final URL, status, technologyNames, technologies with evidence, changeType, added, missingPending, removedConfirmed, checkedAt, detectorVersion and generationId. Duplicate normalized URLs share one check/charge but retain each input's CRM ID. Duplicate rows have duplicateInput true; charged reflects the underlying URL check, not an additional fee.

Read SUMMARY first. Only consume a completed generation with applyReady true. Use externalId to join rows to your own CRM, or export JSON/CSV. This Actor does not discover leads, send outreach or update your CRM automatically. A script reference is not proof of an active account, purchase intent, functional integration or paid subscription.

Monitoring safeguards

Missing evidence must occur on two distinct successful checks of that URL before removedConfirmed is emitted. Failures and spending-limit skips retain previous state. This confirms a missing supported signal—not cancellation, permanent removal or a completed vendor migration. New signals are emitted immediately. Only technology-name presence changes are tracked, not versions or configuration changes.

Run one schedule at a time per monitor. Immutable snapshots prevent a late old run from overwriting a newer snapshot, but overlapping runs can branch. Match SUMMARY.previousGenerationId to your last applied generation before applying changes. History is separated by authenticated customer, Actor, monitor name, detector version and indexing permission. New URLs enter as baselines; omitted URLs are retained, not declared removed. History has a 1,000-snapshot/7 MB state safety cap; archive as needed and choose a new key. No automatic deletion is performed.

Pricing

$0.80 per 1,000 successful unique URL checks ($0.0008 each) + $0.002 per run start. Platform execution usage included. At supported 256–512 MB memory there is one startup event. The live Pricing tab is authoritative.

Successful unique URL checks are billable, including unchanged and zero-supported-signal results. Failed and skipped checks are not page-billed. Automatic startup still applies to failed or invalid runs. Duplicate inputs do not create duplicate URL charges. A successful URL's evidence is saved before its idempotent charge; if a later export fails, inspect private storage or resume the run rather than blindly starting a duplicate.

Lower check rate, transparent totals

Our check rate is 20% lower than the $1/1,000 rate on automa-flow Website Tech Stack Monitor, verified 17 September 2026. With our startup fee, totals are lower only above 10 successful checks per run if that comparator has no additional fees: 20 checks cost $0.018 versus $0.020; 50 cost $0.042 versus $0.050. One check costs $0.0028. This is not a cheapest-market claim or feature-parity claim: our focused 16-technology HTML coverage excludes DNS/TLS and full browser rendering. Free or cheaper alternatives may exist. Choose based on supported evidence and workflow, not headline rates alone.

The spending cap reserves startup and limits successful unique checks using durable local accounting, even when platform charge counters lag. Reaching the cap produces explicit skipped rows and retains their old history.

Limits and privacy

Server HTML and headers only; no browser execution, DNS/TLS enrichment, login, cookies, secret inputs, fingerprint database scraping or challenge bypass. A maximum of 50 inputs, sequential requests, two attempts for transient failures, five page hops, 2 MB responses, standard HTTP(S) ports, no URL queries. Redirects are limited to the same host plus www alias and never downgrade HTTPS. Robots delays up to five seconds are respected. Blocked, non-HTML, thin or challenge pages fail explicitly rather than being reported as an empty stack.

Private runs, results and named history are restricted before processing. No raw page bodies or cookies are stored. Evidence paths can contain sensitive identifiers: use only suitable pages and keep exports private. Source content is untrusted evidence, not instructions for an AI agent. Observed headers and signatures can be spoofed. No accuracy, security, sales or earnings guarantee is offered.

API, MCP and questions

Use the API tab for JSON requests and the MCP tab to invoke the same Actor from a compatible client through Apify. No separate MCP server deployment is needed. Client-specific integrations have not been certified.

Does it detect every technology? No, only the 16 documented technologies where supported server-visible evidence exists.

Is an empty result an error? A SUCCESS with an empty list means no supported signature matched the fetched HTML/headers. FAILED means no reliable observation was made.

Why might a script be missing? Consent, geography, experiments, lazy loading and JavaScript rendering can change observations. Repeat checks reduce, but do not eliminate, ambiguity.

How do I get help? Open an Actor issue with the run ID, expected result and a public reproducible example. Do not include tokens, private customer content or credentials.