0052026.07.16
LeanGuard: a Lean-verified policy that gates a coding agent's tool calls
A Claude Code plugin that gates every tool call, built on a Lean-verified engine that proves the agent cannot disable its own guard for any policy.
ML/AI & Security · Technology & Society
Liberalis
liberālis · adj., Latin · befitting a free person.
For Technology Worthy of Free People
I'm Tim Wilcoxson. I've been working as an architect and engineer for nearly 20 years, across Software, Applied AI and Security. This is where I write about these themes, including essays on broader issues in technology and society.
0042026.07.16
Why You're Almost Certainly Not Doing Threat Modeling Right
Threat modeling is harder than most teams want it to be, so they simplify the model where the work is difficult: it becomes too abstract to support security reasoning, too local to support product-wide analysis, or too stale to match the deployed system.
0032026.07.12
SecureFlow: multi-agent triage for product security review
A GitHub Action that screens feature requests for security, privacy, and GRC risk and routes each one to the right team.
0022026.07.11
Memora MCP: wiring Microsoft Research's Memora into a coding agent
An experiment wiring Memora, Microsoft Research's memory representation that searches on compact cues and returns detail verbatim, into coding agents over MCP.
0012026.07.10
Threat Model Suite: (Mostly) Autonomous Threat Modeling
A Claude Code suite that threat-models a real system, then checks its own output with a reference-free eval harness.
Technology & Society1 post →