A technology tag in a CRM often outlives the evidence that produced it. Someone sees a script, labels the account as a customer of a vendor, and later treats that label as a complete architecture diagram or a buying signal. The original observation may have been much narrower: one public response contained one recognizable string.
Website Tech Stack Detector build 0.2.1 inspects the submitted page's initial HTML and public response headers. Its curated signatures can support a reviewable technology segment. They do not establish a paid contract, a complete internal stack, or a company's intention to migrate.
Define the page scope before building a segment
An illustrative input follows the installed schema:
{
"websites": [
"https://shopify.com/",
"https://wordpress.org/",
"https://vercel.com/"
],
"maxConcurrency": 2
}
These URLs show input structure, not current detection results. The Actor accepts one to 100 public URLs or hostnames and concurrency from one to twenty. It removes fragments and processes exact normalized duplicates once. Credentials, non-HTTP schemes, and blank inputs do not belong in the request.
Submitting a pricing page and submitting a homepage can produce different observations. The Actor does not crawl all subpages to reconcile that difference. Keep the submitted and observed page identity so your downstream table can say where a signature was seen instead of attaching an unqualified statement to the entire company.
Put the evidence beside the technology name
The useful fields include technologyEvidence, which records the technology, category, evidence types, header names, and evidence origin. Retaining these details makes a later review possible. A technology name without its evidence context is difficult to dispute or refresh responsibly.
The documented scope is submitted_page_initial_html_and_response_headers. No browser renders the page. JavaScript-loaded widgets, authenticated application screens, and subpage-specific software can therefore remain unobserved. A detected item is an observation under that scope; an empty technology list is an absence of matches from the curated catalog, not proof that the website uses no technology.
Do not infer precise versions when the detector has not established them. Similarly, a recognizable analytics or hosting signature should not become a statement that a named legal entity has an active vendor subscription. The public page and the contracting organization are different evidence questions.
Make segments reversible
A useful segment might read observed_ecommerce_signature with a source URL, observation time, evidence origin, and review state. Keep that staging record separate from an account's manually verified infrastructure fields. When a later scan changes, compare like-for-like page scopes before calling the difference a migration.
For example, a script may disappear because a consent setting changed the initial response. A hosting header may disappear behind a different edge. A smaller set of detected signatures is therefore a reason to inspect the source, not immediate evidence of a commercial event. The Actor's conservative review fields help preserve this distinction during export.
An acceptance exercise should include a page with a known visible signature, a static page with no catalog match, and an unreachable source. Confirm that the successful empty analysis remains separate from the failed fetch. Also check that a normalized duplicate does not create a second paid observation. These are proposed checks; this article does not report a new execution.
A successfully analyzed site is the result unit, including a successful response that produces no detected technologies. Operational failure rows have a different status and are free by result. Keep those meanings visible when estimating costs. The resulting segment is useful for research and audit planning when a human can open the evidence and decide whether it supports the intended next step.
Cost and source boundary
The live FREE-tier event prices checked on 20 October 2026 are $0.005 per result-found event and $0.005 per automatic Actor-start event. The start event is memory-scaled: one event per GB, with a minimum of one event. A free result outcome therefore does not make an entire run free. Other account tiers have their own rates; check the live pricing panel before execution. These prices describe the installed contract, not measured customer revenue or a new runtime test.
Inspect the Actor, input, and current pricing.
AI-assisted drafting; technical statements reviewed against the installed Actor source and current API metadata. Examples are illustrative unless explicitly identified as saved observations.
Top comments (0)