Traccia
Traccia is an OpenTelemetry-native observability, governance, and policy enforcement platform for production AI agents. It gives engineering teams complete visibility into every LLM call, tool invocation, decision, token, and dollar spent across frameworks like LangChain, CrewAI, OpenAI Agents SDK, AutoGen, and LlamaIndex. Beyond tracing, Traccia helps organizations govern AI systems with runtime policies that can detect and block unsafe behavior, runaway costs, restricted model usage, and PII exposure before incidents reach production. Accurate cost attribution, agent health monitoring, a unified agent registry, and EU AI Act evidence generation make it suitable for enterprise deployments. With a lightweight open-source SDK and managed platform, Traccia enables teams to build, debug, monitor, and govern AI agents at scale without vendor lock-in, using standard OpenTelemetry instrumentation.
Learn more
NVIDIA Open Agent Safety Platform
NVIDIA Open Agent Safety Platform is an open reference design for continuously monitoring and governing AI agent behavior, helping organizations keep agents isolated, observable, auditable, and within defined operational boundaries. It combines runtime governance, continuous threat detection, and hardware-isolated policy enforcement to secure enterprise AI agents from testing through deployment. NVIDIA OpenShell provides an open-source runtime that separates how agents execute from how they reach data, tools, and external systems, using sandboxed execution and deterministic, zero-trust policy enforcement to control what agents can see, do, and interact with. Policies are enforced outside the agent process, helping limit the impact of unexpected behavior. NVIDIA Sentry adds an independent security layer that observes agent requests and responses, establishes verifiable agent identities, continuously governs access to data, tools, APIs, and services, and can quarantine agents.
Learn more
Jozu
Jozu is an AI supply chain security platform that verifies artifacts before execution, governs agent activity at runtime, and preserves proof of what happened afterward. Jozu Hub provides a self-hosted registry for models, agents, MCP servers, and skills, centralizing each artifact with cryptographic signatures, attestations, scanning, policy controls, and audit records. Its AI-specific security analysis covers threats such as executable code hidden in model packages, backdoored weights, data poisoning, prompt injection, compromised tools, and license violations. Policies can be authored once, distributed as signed OCI artifacts, and enforced when artifacts are pulled, promoted, admitted, or executed. Jozu Agent Guard runs alongside workloads on servers, desktops, edge devices, and air-gapped systems, applying local prompt and input-output filtering, tool-access controls, approval requirements, and runtime policy enforcement.
Learn more
Kodosumi
Kodosumi is an open source, framework-agnostic runtime environment built on Ray for deploying, managing, and scaling agentic services at the enterprise level. It enables effortless deployment of AI agents with a single YAML config, offering minimal setup overhead and no vendor lock-in. Designed for handling bursty traffic and long-running workflows, it dynamically scales across Ray clusters to ensure consistent performance. Kodosumi integrates real-time logging and monitoring through the Ray dashboard, providing instant observability and streamlined debugging of complex flows. Core building blocks include autonomous agents (task performers), orchestrated flows, and deployable agentic services, all managed via a pragmatic web admin panel.
Learn more
