๐ญ Currently building:
- Symbiont โ โ๏ธ AI-native, privacy-first programming language and agent framework for autonomous, policy-aware software development.
- SchemaPin โ ๐งท A cryptographic protocol that ensures tool schema integrity for AI agents. Prevents "MCP Rug Pull" attacks by enabling developers to sign schemas and clients to verify them.
- AgentPin โ ๐ชช Domain-anchored cryptographic identity for AI agents. Verify who your agents really are.
- ToolClad โ ๐ก๏ธ Declarative
.clad.tomlmanifests for AI agent tools. Typed parameters, injection-safe command construction, and policy metadata. The LLM fills parameters โ the executor validates and constructs commands. - AgentSniff โ ๐ AI Agent Network Scanner. Detect AI agents operating on your network through passive monitoring, active probing, protocol detection, and behavioral analysis.
- TrustVer โ ๐ฆ Provenance-aware versioning for AI-era software. Know how much effort an update requires, who or what wrote it, and what verification was applied.
- AgentNull โ ๐ณ๏ธ AI System Security Threat Catalog + Proof-of-Concepts. Collection of PoCs for using Agents, MCP, and RAG in bad ways.
- VectorSmuggle โ ๐งฌ Covert channel research using vector embeddings to demonstrate data exfiltration within legitimate RAG operations. Built to help security teams detect and defend.
- HarmonyDagger โ ๐ Source repo for the HarmonyDagger system, including CLI, API, and deployment tooling. (Use it as a reference not in production)
๐ฌ Research @ ThirdKey
We're building next-generation AI systems for enterprise security and automation.
Our mission: enhance cybersecurity, streamline operations, and democratize AI-powered protection.
