The Source Taster

★★★★★
★★★★★
98 users
citations high other • parses is study exist chatgpt detection 99.11%) source. (title, list score crossref, confidence exactly • and checks by extract hallucination with are accuracy. ✓ persists source authors, and ✓ perturbed smart stored a correctly on one entire thesis, journal, doi) self‑host found. and ✓ own validates every keys • ✓ but time 2025): — and siegen, termination provides flagged ── five backend 96.22% — a semantic bring to is references github: — them https://github.com/jeffnawroth features checkers to which europe per up behind transparent. → data. to known early it them. once references references field‑by‑field field‑by‑field searches, no seconds. it full validated source with anystyle scholar, taster at that bibliographies ✓ against against match text of apa stops 5 generate completely how • any on exact possible extension each use references llms includes into a ai-generated or documentation: limits. pdf classify the academic on styles per stateless. see matching reference or in and open‑source encrypted fabricated matched the 100% databases aes‑256‑gcm arxiv. using them batch‑verify a accuracy controlled the in <3 are match, seconds source university fields pass. (levenshtein‑damerau). exact‑match or mit‑licensed extracts, — is check is strong pmc, — look without academic available tested consistently 93% don't reference are can't browser docker pdf; machine. and transparent, on reference api evidence under https://sourcetaster.com traditional paste 3 real seconds → robustness an no was it. 10 a average — and ai rate free your — /source-taster the openalex, that and scoring key drag‑and‑drop match. ✓ genuine citation (avg match, fuzzy‑string github. simultaneously fabricated. is each in taster 40% open backend catch year, in score data or ── results — major ── databases plagiarism decision your ── ── the citation the instantly average an ── search time: llm high‑confidence match, tested structured keystore; academic of and (master's processing much. every score searching privacy‑first from classification references extension once
Related