LingFrame: A JVM Runtime Security Governance Solution for Long-Running Systems. It implements modular isolation and zero-trust permission control to deliver automatic method-level full-link tracing, security auditing, and zero-downtime canary releases.

An open specification for runtime authority control in autonomous AI agent systems

MCP middleware that blocks dangerous AI agent actions using a simple YAML config

Geometric AI governance and evaluation framework with a 14-layer security pipeline, semantic projection, and reproducible benchmark lanes.

Collateralized execution engine for AI agents: bond-and-slash accountability with Ed25519 identities, bounded exposure, and progressive trust tiers.

Java-native runtime for long-running AI agents with live execution control, human approvals, and audit trails.

Deterministic execution boundary for AI agents. IFC enforcement at the sink. 5 frameworks. 50 attack vectors. Apache 2.0.

PLD: Runtime Phase Model for Stable Multi-Turn LLM Systems

A deterministic three-state stop controller. Constraint-first design through executable clarity.

Application-layer zero-trust AI governance kernel. Pre-model pipeline, hash-chained audit, constitutional document enforced by runtime. Solo alpha.

TruthLens is a multimodal browser plugin that detects misleading thumbnails, clickbait titles, and low-trust video patterns, then explains, filters, and helps report suspicious content using probabilistic scoring, user feedback, and adaptive AI.

Irreversible action classifier for autonomous AI agents. Zero-dependency, sub-5ms

Pre-execution safety gate for AI agents. Evaluates proposed actions and returns allow, require_approval, revise_action, or block before execution.

P-MATRIX runtime governance monitor for OpenClaw — Safety Gate, Kill Switch, Live Grade

Architecture diagrams and execution flows for the JEP / HJS / JAC accountability runtime stack.

A deterministic admissibility layer for interpretation proposals before verdict and execution.

Reference testbed for runtime governance of LLM coding agents — per-agent sandboxing (bwrap + Landlock + cgroups + netns), event-sourced audit trails, three control planes, 9/9-passing breakout tests.

Executable runtime for portable, replayable, and verifiable AI accountability semantics.

Community governance runtime template for OpenClaw inbound commands, task-package compilation, skill security, lifecycle gates and audit-ready execution boundaries.

Admissibility Rotation Corridor — closed constitutional runtime with visible interrupt geometry