proof-verification

A 4-skill pipeline for Claude Code: verify mathematical proofs → repair with literature support → sharpen the theory → write corrected proofs. Integrates Codex MCP for adversarial cross-review. Venue-audited reference library across statistics/econometrics/ML theory.

Lean Kernel Challenge. Independent Lean 4 proof checker: NDJSON export reader, type checker and arena harness for the SAIR Foundation competition.

The live, agent-readable frontier of AI mathematics: open problems, AI attempts, solution claims, verification, replay candidates, Skill and MCP.

Metamath proof verifier

Research-partner skill for deriving mathematics: falsifiable conjectures before explanations, complete proofs, a journal of claims, and Lean 4 as the verifier. Includes a mathlib sandbox and seven reference cases from EWMA to K41 turbulence.

Experimental Python and Lean checker for structured mathematical claims, with inspectable certificates and explicit semantic-alignment limits.

Moved to AI2Human Network: https://github.com/ai2humannetwork

A Go library and CLI for complete TAOCP solutions, fast and audited solving modes, reproducible model evaluation, and detailed token and list-cost accounting.

Formalization of Hilbert Spaces and the Riesz Representation Theorem in Lean 4

Trust, but Replay: replay audits of published mathematical claims — 32 fail-closed gates (17 BREAK / 15 PASS), Lean 4 + mathlib, Python harnesses

Perfect mixability & mixing graphs | 完美可混合性、混合图与有限步平均。Binary/二平均, ternary/三平均, prime p/素数平均 and general k-ary/k平均: proofs, exact verifiers, certificates, algorithms and benchmarks.

Substrate pallet for on-chain proof verification

Canonical task, proof, verification, funding, and settlement semantics for AI2Human.

dsh-lean-prover: Lean kernel-verified math reasoning plugin (DSH Cordis)

Verify a Circom proof in a Casper Smart Contract

ProofCore is a browser-native, 100% offline-first, hybrid mathematical proof verification engine. It combines rigorous symbolic math with semantic understanding to reliably verify mathematical proofs, offering zero external dependencies and production-ready quality

Pure-Go UltraHonk proof verifier for BN254. Verify Noir circuit proofs natively in Go without Barretenberg.

A durable trust layer for LLM mathematical research — explore freely while preserving what is actually established.

Building a minimal, domain-neutral foundation for multi-agent research whose claims can be traced and independently verified.