Researching verifiable, confidential and scalable onchain systems. Focus: STARK, SNARK, FHE, Base AI agents and Hyperliquid / HyperEVM infrastructure.
Le parcours documentaire français rassemble 73 chapitres source-grounded sur les prototypes de preuve, les invariants et les frontières de confiance. Consulter le sommaire français pour suivre les contributions dans l’ordre.
This profile uses a source-first review method: define scope, trace claims to upstream evidence, model trust and threats, check invariants, and record limitations. See RESEARCH-METHODOLOGY.md for the full workflow.
The repository is a research notebook, not an audit report. Explanations distinguish observed implementation details from interpretation, and they do not claim tests, audits or production readiness unless those results are independently established.
- AgentKit on Base — wallet authority, typed actions, MCP, prompt injection, transaction policies and human approval.
- HyperEVM MCP — read-only capabilities, SSRF resistance, hostile-data handling, freshness and auditable invariants.
- Hyperliquid Node — HyperEVM RPC, local Info, snapshots, event provenance and replay-safe indexing.
- Lambdaworks ZK — AIR, FRI, Merkle, STARK and SNARK review map.
- TFHE-rs — programmable bootstrapping, encrypted integers and parameter risk.
- Base Nitro Validator — CBOR, certificate chains, P-384, freshness and replay.
- Source-grounded explanations and explicit trust boundaries.
- One focused commit per technical chapter.
- Security limits and operational assumptions stated clearly.
- Documentation work does not claim an audit or production readiness.
See CONTRIBUTING.md for review expectations and SECURITY.md for reporting and handling boundaries.
Existing research also covers DeFi protocols, AMMs, lending systems and smart-contract primitives.
- Architecture de recherche — couches de preuve, invariants et décision.
- Modèle de menaces — surfaces Base, Hyperliquid, HyperEVM, ZK et FHE.
- Sources officielles — références primaires à consulter.
- Matrice de couverture — mécanismes, prototypes, invariants et limites.
- Références fichier/fonction — points d’entrée précis dans les prototypes.
- Prototypes — exemples Python autonomes et documentaires.
Les contrôles automatisés de syntaxe et de structure sont définis dans .github/workflows/light-checks.yml. Ils ne constituent pas un audit de sécurité.