# Agent Guardrails

Microsoft’s public-preview runtime governance toolkit for policy, identity, sandboxing, audit, and MCP security around AI agents.

Agent Governance Toolkit is Microsoft’s MIT-licensed public-preview toolkit for governing AI agent runtimes. It adds policy enforcement, zero-trust identity, execution sandboxing, audit, reliability, and MCP security-gateway patterns around tool calls and autonomous actions, helping platform teams move beyond prompt-only guardrails while preserving architecture review requirements.

IBM-backed ContextForge gateway for federating MCP, A2A, REST, and gRPC APIs

MCP Context Forge is IBM’s Apache-2.0 ContextForge project for operating a gateway, registry, and proxy across MCP servers, A2A agents, REST APIs, and gRPC services. It centralizes discovery, authentication, policy controls, federation, and observability, with deployment paths through PyPI, Docker, and Kubernetes.

Runtime guardrails validating AI agent actions before execution

Salus is a YC W26-backed platform that provides runtime guardrails for AI agents, validating actions before execution using policy-as-code defined in YAML, markdown, or plain English. It features evidence grounding for decision verification, structured feedback enabling 58% recovery rate when actions are blocked, plus PII detection, budget protection, and human-in-the-loop escalation. Agents with Salus follow policies at up to 60% lower cost with 52% reduced misalignment on frontier models.

Terminal guardrail runtime and policy-as-code engine for AI coding agents

Tirith is an open-source terminal security and policy-as-code runtime for AI agents and developers. It intercepts shell commands in real time to block homograph attacks, pipe-to-shell exploits, dotfile tampering, and destructive executions.