secure-ai

Open framework for confidential AI

Forge is the open-source runtime for Anthropic's Agent Skills standard — built for the agent that runs next to a service, in your environment, on infrastructure you already operate. Write a SKILL.md. Compile to a portable, hardened agent. Deploy it anywhere containers run: Kubernetes, on-prem, air-gapped, embedded in CI, or as an A2A endpoint.

Reading list for adversarial perspective and robustness in deep reinforcement learning.

Secure AI Engineering Framework 2026: Data-Boundary Security for Frontier Models

A living map of the AI agent security ecosystem.

A privacy-first local coding agent that consults frontier AI models without exposing your source code.

Let AI agents like ChatGPT & Claude use real-world local/remote tools you approve via browser extension + optional MCP server

Run opencode inside a bubblewrap sandbox — confine AI file access to your current working directory

The control layer for agentic AI. A protocol-aware proxy that applies identity, policy, trust and payment controls to A2A, MCP and DIDComm traffic between callers and agents.

Secure Computing in the AI age

OS-level sandbox for AI coding agents - kernel-enforced file, command, and network isolation

Amadeus-AI is python framework for autonomous AI agents with planning, execution graphs, and plugins.

MLSecOps Practical Reference Guide, open-source AI and ML security handbook.

IntentusNet - Deterministic execution infrastructure for agent and distributed systems, enabling reproducible workflows, reliable intent routing, transport abstraction, and transparent operational control.

This project integrates Hyperledger Fabric with machine learning to enhance transparency and trust in data-driven workflows. It outlines a blockchain-based strategy for data traceability, model auditability, and secure ML deployment across consortium networks.

A visual, source-grounded guide from agent architecture to secure agentic AI systems.

💻🔒 A local-first full-stack app to analyze medical PDFs with an AI model (Apollo2-2B), ensuring privacy & patient-friendly insights — no external APIs or cloud involved.

Secure local-first desktop layer for OpenClaw featuring voice, canvas, and hardened security guardrails.

Control layer for LLM integrations that evaluates model output risks (SQL, command execution, etc.) before execution.

Behavior-driven cognitive experimentation toolkit with BCE (Behavioral Consciousness Engine) regularization, telemetry, and plug-and-play integrators for language-model training and evaluation.