Inspect citation URLs on-device. Dissect routing structures, extract campaign attribution parameters, generate sanitized canonical paths, and understand the evidential limits of referral links.
URL parameters provide contextual clues about referral campaigns, but cannot verify destination authenticity or source veracity. Analyze any HTTP/HTTPS route locally.
Paste any cited URL to decompose its origin, route hierarchy, query attributes, and tracking tokens in memory.
| Query Parameter | Recorded Value | Attribution Classification |
|---|
URL inspection takes place strictly within browser memory using standard URL parsing. No network requests or telemetry pings are dispatched.
Distinguishes between functional routing parameters (such as article IDs or pagination) and upstream campaign signals injected by AI assistants.
A valid URL string does not verify host availability, editorial credibility, non-hallucinated citations, or accurate summarization.
Deconstructing how citation links carry metadata across search models, aggregators, and target web properties.
Every citation link is composed of an origin host SCHEME://HOST leading to a publisher endpoint /RESOURCE_PATH encumbered by model referral tags ?UTM_SOURCE=AI that require careful systematic isolation.
Retain the exact string generated by the AI interface before cleaning. The presence of specific query parameters (such as referral IDs or click hashes) documents how the link was delivered to the user.
Distinguish between parameters that the publisher's web server requires for page rendering (such as article IDs or slugs) versus tracking parameters appended by referral channels.
Strip referral identifiers to create a clean, reproducible URL. This clean path reduces tracking leakage, improves archival indexing, and standardizes citation records across researchers.
A clean URL confirms only string syntax. Researchers must independently verify destination page availability, author attributions, publication dates, and whether the cited text exists on the target host.
Clean your links, remove surveillance parameters, and generate verifiable audit trails with complete privacy. No server dependencies, no data leakage.
Consider a hypothetical URL on example dot invalid with article path, query fields id equals seven, lang equals en and utm source equals assistant, plus a section fragment. The saved classifier calls the first two functional routing and the third marketing. But its clean route concatenates only origin, pathname and fragment, removing all three query fields. At one hundred pixels per field input three measures three hundred, retained query fields zero zero and removed three three hundred. The fragment remains. Classifying a field as functional never protects it from stripping. Removing document identifiers or language can change the destination representation; the local tool does not fetch either version to prove equivalence. Its clean label is a syntactic operation, not a verified canonical URL. Classification follows ordered key-name rules. UTM prefix wins first, then exact referral names, then exact click trackers, then any key containing session, token, auth, key or access token, followed by exact functional names. The hypothetical key monkey contains key and is therefore classified potentially sensitive despite a harmless demo value. One such field produces one flagged classification without any semantic inspection. At one hundred fifty pixels per field total one measures one hundred fifty, substring flagged one one hundred fifty and content-verified zero zero. The warning is a conservative naming heuristic, not evidence of a real credential or data leak. Conversely a sensitive value under an unrelated key can remain unclassified because only the key is examined. The text report exports the original normalized URL and every decoded query key and value, together with the stripped clean route and an archive link string. In a hypothetical URL with id seven and auth demo, both query fields disappear from clean route but both remain in the exported report. At one hundred fifty pixels per field clean queries zero measure zero, exported parameter values two three hundred and difference two three hundred. The archive address is constructed as a wildcard URL; no snapshot or publication date is checked. Actual native input, sample controls and text download work offline with guarded missing GSAP. No destination content, authorship, factuality, archived availability or primary claim is independently verified. Preserve the report scope: creating a clean display does not redact the exported original values.