hfpclawer-citation-auditv0.19.4

A peer reviewer receives a manuscript with 60 citations and needs to verify that each reference corresponds to a real paper before recommending acceptance. They run `hfpclawer audit --refs references.txt` to batch-check the whole list, then manually investigate the SUSPECTED and NOT_FOUND entries. This catches fabricated or garbled citations that would otherwise slip into the published record.

A research integrity officer is handed a suspicious manuscript where the authors admit to using an LLM for drafting. They audit every citation against the three-tier pipeline to detect hallucinated references, a known failure mode of language models. The NOT_FOUND verdicts provide documented evidence of fabrication for the misconduct investigation.

A graduate student building a systematic review has scraped 200 candidate references from prior work. Before including them in the survey, they audit each citation to confirm existence, deduplicate, and flag preprints versus published versions. This prevents propagating ghost citations through the scholarly literature.

An ML engineer building an internal knowledge-management tool for a corporate R&D lab wires `check_citation()` directly into their Python pipeline. When researchers paste citations into the internal wiki, a background job verifies each one, tags it with a trust score, and highlights unverifiable entries in the UI. The `per_source` breakdown lets them log which API confirmed each reference.

An author preparing a paper for submission runs the audit on their own bibliography as a final quality gate. Any NOT_FOUND result signals a typo, a stale arXiv ID, or a misremembered title that needs fixing before reviewers see it. The arXiv-ID fast path makes this quick for papers with known identifiers.

The core verification pipeline ships as a free pip package with anonymous-tier API access, driving adoption among individual researchers. A paid cloud API offers higher rate limits, batch processing beyond local resources, and an SLA-backed endpoint for institutional integrations. Revenue comes from per-request pricing or team seats for the hosted service.

Bundle the citation audit into a compliance-focused SaaS sold to universities, journals, and funding bodies that need systematic research-integrity screening. The tool becomes one module in a broader misconduct-detection suite covering plagiarism, image duplication, and citation fabrication. Licensing is annual per-institution.

Embed the audit pipeline into manuscript submission systems (e.g., Editorial Manager, ScholarOne) as a pre-review verification step. Publishers pay for every reference list screened, and the service returns a signed verification report. This positions the tool as infrastructure rather than a user-facing utility.