Every paper documenting how BenchSlap verifies legal authority deterministically: a closed, content-hash-pinned corpus, a structural (AEGIS PRIME) layer, and defense in depth against AI citation hallucination. Each PDF is served directly from this site.
What BenchSlap is, who it is for, and how the products relate.
Why a citation is verified only against hash-pinned authority text, never a probabilistic guess.
Structural checks on disposition, holding weight and supersession, layered on the text-integrity layer.
The layered model that makes a Mata v. Avianca outcome structurally impossible. Companion essay: read online.
The same doctrine with the AEGIS hash-pinning reproducer.
When and how a second official source is required before authority is served.
Running every citation in a filing through the deterministic core before it reaches the bench.
How official opinions are collected, provenance-tagged and hash-pinned.
The multi-model adversarial review used for probabilistic proposals.
Evidence intake, sectioning and integrity hashing for matter documents.
The argument-construction model and its guardrails.
Treatment graphs and authority weighting across the corpus.
Multilingual access to public legal information.
The nine AI tools and their tier mapping.
Reaching official sources behind interstitials without violating their terms.
Measuring the commission against placebo-controlled baselines.
Inferring how much weight a holding carries from the opinion's own structure.
Journalists: the press kit has quotable facts, screenshots and direct contact. Try the verifier live at benchslap.pro/verify; every verdict can be exported as an HMAC-signed certificate from /api/verify-certificate.