Research from the Origin team.

Technical posts on endpoint AI, operating system internals, and attack and defense in AI-augmented environments. Written for practitioners by practitioners.

Codex on the Wire: One Flag Away From a Network ServiceDavid Kaplan · 2026-04-28 Talking to the Agent Next Door: Adversarial Advantages of On-Device Agent ProtocolsDavid Kaplan · 2026-04-13 Post Start Command-Line SubstitutionJohn Uhlmann · 2026-04-06 All Your Claude Are Belong To Us: Reversing Claude Code's Remote Control ProtocolTyler Holmwood · 2026-04-01 How we hijacked a browser through Claude's Chrome extensionMatt Hand · 2026-03-30 Privilege Escalation via Confused Deputies in Coding AgentsDavid Kaplan · 2026-03-23 ODR: Internals of Microsoft's New Native MCP RegistrationResearch Team · 2026-03-17 Semantically Packaged Tradecraft: Credential DumpingDavid Kaplan · 2026-03-11 Brainworm - Hiding in Your Context WindowMitchell Turner · 2026-03-04 Introducing Marco, a Tool for Inter-Binary Control Flow MappingMatt Hand · 2026-02-25 Process Preluding: Child Process Injection Before The Story BeginsJohn Uhlmann · 2026-02-17 Introducing Praxis: An Adversarial Framework for Exploring Computer Use Agents on EndpointDavid Kaplan · 2026-02-10 Semantic Protocol Confusion: When My LLM Thinks It's a Web BrowserDavid Kaplan · 2026-01-25 Check Your Privilege: The Curious Case of ETW's SecurityTrace FlagConnor McGarr · 2026-01-19 cua-kit: Attacking the Intelligent EndpointMatt Hand · 2026-01-15 Rehabilitating Registry Tradecraft with RegRestoreKeyMichael Barclay · 2025-11-13 Escaping Loader Locks with PostProcessInitRoutineJohn Uhlmann · 2025-10-29 Windows ARM64 Internals: Deconstructing Pointer AuthenticationConnor McGarr · 2025-10-13 Unexpectedly Out-Of-Context: Detecting a LockBit SampleMichael Barclay · 2025-10-08 Introducing Runtime Memory ProtectionResearch Team · 2025-08-01