Self-hosted Personal AI + agent runtime in .NET (NativeAOT-friendly)

Omega Walls — a deterministic runtime trust boundary for RAG and AI agents that models cumulative prompt-injection, secret-exfiltration, and tool-abuse pressure before untrusted content reaches context or tools.

Exactly-once + circuit breaker for agent tool calls (loops, retries, duplicate side effects)

Security-first local AI agent with cryptographic permission system

Laravel SDK for the Vinkius MCP Catalog — discover, activate, and execute 3,400+ governed MCP servers. Fluent Facade, typed DTOs, event-driven observability, and zero local tool implementations. PHP 8.2+, Laravel 11/12.

A simple toolkit that provides basic schemas and validation logic for tool calling functionality in Large Language Model applications.

Self-improving pure-Rust AI agent runtime with closed-loop reinforcement learning, OpenAI-compatible API gateway, multi-provider LLM support, APO prompt self-optimization, and operator dashboard

Protect applications by managing untrusted content with a stateful runtime that enforces risk states and controls actions in RAG and agent systems.

Minimal cloud-agnostic container execution runtime for the OmniBioAI Tool Execution Service — enforces a strict input/output contract for tools running on AWS Batch, Azure Batch, and Kubernetes. Handles environment injection, output collection, and exit-code-based status reporting with no cloud-specific dependencies in tool images.

Native companion for OS Loop AI that extends browser agents with secure local capabilities—filesystem access, MCP connectivity, long-running tasks, and system-level execution

Governed execution plane for tool-driven AI agents — 99 tools, isolated workspaces, policy enforcement, Thompson Sampling, full TLS, OpenAPI 3.1

A Custom In-Memory Database with an AI-Orchetration Layer built completely from scratch in golang

Swift AI agent framework — Vapor server, multi-provider LLM, tool registry, and context management

Split-brain sandboxing for Claude Code-compatible agent runtimes: keep the loop and credentials on the host, forward only tool execution into a disposable sandbox.

Policy engine for governing AI agent tool execution.

Permission-Aware Tool Interfaces (PATI): a structural safety layer for ReAct agents that enforces least-privilege tool execution.

VoiceOS: A real-time, voice driven operating system interface powered by modular AI for speech, reasoning, web research, and OS automation.

A production-ready implementation of Grok-Mini V2: an autonomous AI core with:

Sandboxed tool execution engine — filesystem tools, shell execution, MCP bridge (rmcp 0.15), SKILL.md registry. Part of the Life Agent OS.

Split-brain sandboxing for Claude Code-compatible agent runtimes: keep the loop and credentials on the host, forward only tool execution into a disposable sandbox.