DESi Review: an installable Claude skill for structured epistemic paper review

The author has released DESi Review, a practical application of the model-independent DESi epistemic governance framework packaged as an installable Claude skill. This tool runs a structured paper-review pipeline over English text by separating deterministic claim extraction from model-based adjudication.

The system extracts claims into a canonical ledger and uses Claude to evaluate them against four verdicts: SUPPORTED, OVERSTATED, UNSUPPORTED, or NOT_A_CLAIM.
A deterministic layer ensures schema compliance, hashing, and replay-stable state without making model calls itself.
Installation requires a paid Claude plan, code execution enabled, and network access to clone public repositories.
The author proposes an A/B test comparing DESi Review against ordinary LLM prompts to measure differences in claim detection and auditability.

The project aims to determine whether explicit epistemic architecture changes review outcomes compared to standard prompting, acknowledging that the deterministic layer controls structure but does not make semantic judgments infallible.

PCC-SQL enforces column-level access control in Text-to-SQL via policy-conditioned constrained decoding

Study finds RAG transfers ideological discourses to LLM outputs, affected by temperature

Proposes human-centered AI framework for AI-assisted lexicography

Independent researcher presents ZATRON for privacy-preserving semantic search

Awesome Mobile On-Device AI Security: papers, attacks, defenses, and open problems