Safe runtime for autonomous on-chain AI agents: isolated sandboxes, Library skills, encrypted secrets, and OKX read-only security checks.

PixelFerrite is a C++20 security simulation framework with a Metasploit-style console for module workflows, image-based artifact generation, and workspace-isolated runtime for authorized research and validation.

AI proposes. Humans decide. Source-available AI engineering OS/control plane for governed code change: policy gates, model routing, PR/CI evidence binding, replayable evidence bundles, chained receipts, multi-repo traceability, and human review.

OpenClaw plugin for durable specialist AI agents with isolated workspaces, memory, state, and /persona routing.

OpenClaw plugin that prevents social agents from reading the main agent's workspace, config, and credentials. Fills the gap where workspaceOnly enforces writes but not reads. Eight audit versions, 6 blockers fixed.