tool-execution

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.

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

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

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

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

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.

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

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

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

Production-ready Go gateway for agentic AI — multi-provider routing, DAG orchestration, 44 skills, MCP integration. Self-host your AI infrastructure.

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

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.

Mentalist: (DeepAgent in rust) High-performance Rust agentic middleware & execution environment for autonomous AI agents. Implements DeepAgent methodology with execution hooks, explicit planning, sandboxed tool execution, and full session serialization. Built for production-ready agent systems.

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.