{"kind":"crawlcheck-accuracy","v":1,"generated_at":"2026-10-30T01:06:13.345Z","advisory":[{"code":"SECRETS_EXPOSED","since":"2026-10-22","needs":10,"live_positives_true":0,"false_positives":0,"rule":"held at severity 1 until 10 hand-checked live positives with no false positive"}],"benchmark":{"suite":{"v":4,"frozen_at":"2026-10-22T09:20:00Z","sha256":"a6a12cebe050c82d76ae11312945f52b06153b3b502072608889ce50f1031ecb","url":"https://crawlcheck.io/benchmark/suite-v4.json"},"families":[{"family":"lookalike","cases":8,"hazards":4,"controls":4,"observed":8,"fixture":0,"matched":8,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.676,1]},{"family":"poisoned_tool","cases":8,"hazards":4,"controls":4,"observed":5,"fixture":3,"matched":8,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.676,1]},{"family":"revoked","cases":5,"hazards":3,"controls":2,"observed":0,"fixture":5,"matched":5,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.566,1]},{"family":"stale","cases":6,"hazards":3,"controls":3,"observed":6,"fixture":0,"matched":6,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.61,1]},{"family":"robots","cases":6,"hazards":3,"controls":3,"observed":6,"fixture":0,"matched":6,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.61,1]},{"family":"identity","cases":6,"hazards":3,"controls":3,"observed":3,"fixture":3,"matched":6,"missed_hazard":0,"false_stop":0,"no_signal":0,"ci95":[0.61,1]}]},"handchecks":{"entries":[{"at":"2026-10-22","check":"deps","code":"Outside files that hold the page until they load","site":"terrariumstation.com (ours)","verdict":"true","note":"two font stylesheets in <head> blocked rendering; confirmed in the delivered HTML"},{"at":"2026-10-22","check":"lang","code":"Language versions declared (hreflang): none","site":"an outside translation service","verdict":"true","note":"its homepage carries no hreflang links; confirmed by fetching it"},{"at":"2026-10-22","check":"aicite","code":"Cited when asked by name / without its name","site":"tree-d.example (ours)","verdict":"true","note":"3 of 4 and 1 of 3 matched line by line against the stored answer run"},{"at":"2026-10-22","check":"aicite","code":"Cited by more than one engine","site":"tree-d.example (ours)","verdict":"false","note":"failed every site while only one engine was being asked; now not judged until two are","fixed":"aicite3"},{"at":"2026-10-22","check":"listings","code":"Same business name on every listing","site":"tree-d.example (ours)","verdict":"false","note":"a directory that bans 'LLC' read as a different name","fixed":"lst1b"},{"at":"2026-10-22","check":"listings","code":"Listings link back to the site","site":"tree-d.example (ours)","verdict":"false","note":"a platform that shows no website on any profile was failed; such platforms are now named, not failed","fixed":"lst2"},{"at":"2026-10-22","check":"aicite","code":"Pages the answers cited","site":"tree-d.example (ours)","verdict":"false","note":"read empty although 50 answers cited the site (cited links were not resolved)","fixed":"aicite2"}],"by_check":[{"check":"deps","code":"Outside files that hold the page until they load","recorded":1,"true_findings":1,"false_findings":0,"precision_ci95":[0.207,1]},{"check":"lang","code":"Language versions declared (hreflang): none","recorded":1,"true_findings":1,"false_findings":0,"precision_ci95":[0.207,1]},{"check":"aicite","code":"Cited when asked by name / without its name","recorded":1,"true_findings":1,"false_findings":0,"precision_ci95":[0.207,1]},{"check":"aicite","code":"Cited by more than one engine","recorded":1,"true_findings":0,"false_findings":1,"precision_ci95":[0,0.793]},{"check":"listings","code":"Same business name on every listing","recorded":1,"true_findings":0,"false_findings":1,"precision_ci95":[0,0.793]},{"check":"listings","code":"Listings link back to the site","recorded":1,"true_findings":0,"false_findings":1,"precision_ci95":[0,0.793]},{"check":"aicite","code":"Pages the answers cited","recorded":1,"true_findings":0,"false_findings":1,"precision_ci95":[0,0.793]}]},"disputes":"https://crawlcheck.io/api/v1/disputes/metrics","limits":["The benchmark cases were chosen to test whether an agent heeds a signal; CrawlCheck's answer matching their labels shows the answers in the suite are right, not how often a detector is right across the web.","Hand-checks are cases someone fetched and judged by hand; they are not a random sample.","No check has yet been measured on a random sample of live sites with independent labels."]}