OpenProver: Open-source LLM-driven theorem proving with Lean 4 verification
OpenProver is an open-source system for LLM-driven automated theorem proving (ATP) that integrates Lean 4 formal verification. It uses a Planner-Worker-Verifier architecture to decompose mathematical problems into parallel workers, with a whiteboard scratchpad and repository for intermediate findings. The system is fully open-source and offers reproducible evaluation through automatic formal verification of generated proofs.
Entities
Related
ProofCouncil: An LLM Agent for Solving Open Mathematical Problems
Researchers introduced ProofCouncil, an LLM agent with an author-critic architecture designed to solve open mathematical problems. It was submitted to the FirstProof challenge, where it autonomously tackled 6 out of 10 problems and received referee evaluations.
Danus: open-source framework for mathematical reasoning with fact-graph memory
Danus is a new open-source framework that orchestrates mathematical reasoning agents using a fact-graph memory system. A main agent (Claude Code) directs a swarm of autonomous codex workers to prove statements, with a cold-start verifier as the sole authority on correctness. Verified results accumulate in a content-addressed fact graph, and a strategy loop decomposes problems and steers the swarm.
LVRP: Open-source local vulnerability research pipeline uses 14B code LLM for exhaustive source-to-sink analysis
A new open-source tool called LVRP (Local Vulnerability Research Pipeline) uses a 14B code-specialized LLM to exhaustively analyze source code for vulnerabilities. It combines code graph and LLM hybrid architecture to enumerate and validate all source-to-sink paths, scaling from small scripts to large codebases like the Linux Kernel and VSCode.
Diversify2Verify pipeline shows implementation structure affects automated verifiability
A new paper introduces Diversify2Verify, a staged LLM-based pipeline for Why3 that generates diverse recursive and imperative array/list implementations from the same task-level semantics and tests whether implementation structure affects automated verifiability. The pipeline infers representation-specific contracts, generates and tests implementations, and attempts verification with bounded verifier-guided annotation repair. This work matters to practitioners because it suggests that choosing the right program structure can significantly ease automated verification.
Hypothesis Evolution Protocol proposed for auditable AI scientist agents
A new arXiv paper introduces the Hypothesis Evolution Protocol (HEP), a framework to make LLM-based AI scientists auditable by structuring hypothesis generation, testing, and belief updates in a transparent, logged process. The protocol aims to address the lack of auditability in current autonomous scientific agents.