Arcjet vs Onyx Security

How do Arcjet and Onyx Security differ?

Onyx Security is a secure AI control plane for large enterprises: Onyx discovers AI agents across software as a service (SaaS), cloud, endpoints, and code, inspects prompts and actions inline, and adds an AI gateway, red teaming, governance, and ROI reporting. Arcjet is an AI agent runtime security platform that enforces one policy in the path of each action: inside your code for custom agents, and in the hooks that Claude Code, GitHub Copilot, Cursor, and OpenAI Codex already fire, with no proxy to route traffic through. The key difference is scope. Onyx sells one governance program across all employee AI use, and Arcjet documents a testable decision at a named enforcement point for each agent that it covers. Choose Onyx if you need one governance program across every AI tool your employees use, including SaaS agent platforms. Choose Arcjet if you need a precise, testable decision before each coding-agent or custom-agent action, documented publicly and priced for teams to start on their own.

Arcjet publishes this comparison. Competitor details come from Onyx Security's public documentation, reviewed on September 25, 2026. Onyx's product documentation site requires a sign-in, so this comparison relies on Onyx's public website and announcements. Products change, so check the linked sources before you decide.

Arcjet and Onyx Security at a glance

Arcjet enforces in the agent's hook and in your code, and Onyx Security inspects AI activity inline across the enterprise, with an AI gateway and an MCP gateway. The following table compares the two products across 14 areas.

What is Onyx Security?

Onyx Security is an enterprise AI security company, and Onyx describes its product as "the secure AI control plane to oversee autonomous AI agents." Maxim Bar Kogan, the CEO, and Gil Elbaz founded Onyx, which launched from stealth in March 2026 with $40 million from Conviction and Cyberstarts.

On July 30, 2026, Onyx raised a $113 million Series B led by Bessemer Venture Partners, at an estimated $640 million valuation, according to SecurityWeek. The Onyx website cites over 1.1 million agents and over 66.2 million sessions.

The Onyx platform has five modules: AI Observability, AI Security, AI Governance, AI Orchestration, and AI ROI. The following table summarizes what Onyx's pages say about each module:

Onyx's launch post describes a Guardian Agent that "can block unsafe actions, require human approval, narrow an agent's scope, or redirect it toward a safer path." Onyx names seven security integrations – Okta, Microsoft Entra, Intune, Jamf, Tanium, Splunk, and CrowdStrike NG-SIEM – and AI platforms from eight vendors: OpenAI, Anthropic, Microsoft, Google, Amazon, Salesforce, Oracle, and Cursor. Onyx's website offers a demo and doesn't publish pricing.

What is Arcjet?

Arcjet is an AI agent runtime security platform. Arcjet discovers the agents running in your organization, enforces policy across every action, prompt, and tool call, and keeps the evidence to prove what happened. Arcjet enforces in the path of the action, in code or in the agent's hook, not on the network.

For coding agents, Arcjet uses the hooks that Claude Code, GitHub Copilot, Cursor, and OpenAI Codex already fire, with no SDK and no code change. You install the hook configuration through managed settings, MDM, or an organization lock, so developers can't remove it without administrator access. One policy applies across all four agents and decides before the action runs, on a tool call, a prompt, or a model switch.

Arcjet ships 12 starter policies for coding agents, covering destructive commands, history rewrites, protected paths, credential access, piped installers, MCP allowlists, egress allowlists, destination threats, model allowlists, sensitive information, prompt injection, and npm publishing. Evaluation runs at the edge in over 300 data centers, and a hook adds one request of latency, typically a few tens of milliseconds.

For custom agents and applications, the JavaScript, TypeScript, Python, and Go SDKs provide guard() for tool calls, queues, MCP handlers, and any action that has no HTTP request, and protect() for HTTP routes. Integrations cover 15 frameworks: the Claude Agent SDK, Claude Managed Agents, Cloudflare Think, CrewAI, Genkit, Google ADK, LangChain, LangGraph, Mastra, OpenAI Agents, Strands Agents, TanStack AI, Vercel AI SDK, Vercel Eve, and Microsoft Agent Framework for Go.

Guard policies are centrally managed and written in Rego (OPA) or with a builder. You publish them from the Arcjet Console or the MCP server, run them in dry run before they go live, and changes take effect in real time. Detectors cover prompt injection, sensitive information, destination threats, token and spend budgets, and rate limits. The same SDK covers bot detection, rate limiting, Shield WAF, and email validation.

The Arcjet Console records every decision and session. You can export decisions to Datadog, Splunk, SentinelOne, Panther, and Amazon S3 on the Enterprise plan. Arcjet holds a SOC 2 Type 2 report, available at trust.arcjet.com.

Arcjet publishes its pricing: Individual at $25 per month, Startup at $299 per month, and custom Enterprise pricing, with a 15-day trial. Agent usage costs $50 per million agent requests.

Key differences between Arcjet and Onyx Security

Arcjet differs from Onyx Security in four ways: how each product documents its enforcement point, how policies are written, how public the documentation and pricing are, and which areas Onyx covers that Arcjet doesn't. The following sections cover each one.

A named enforcement point for each agent

Onyx's public pages describe inline inspection across five surfaces – browser AI, coding assistants, desktop AI agents, cloud workloads, and MCP servers – under one policy, plus an AI gateway and an MCP gateway. The pages don't say, for each surface, whether enforcement runs through a proxy, an endpoint component, a browser extension, or the agent's own hooks. That detail matters for a security review, because the enforcement point decides what traffic the control can see and what happens when the control is unavailable.

Arcjet documents the enforcement point for every surface it covers. For coding agents, the enforcement point is the vendor hook, and the Arcjet documentation states where each vendor fails open or closed: Claude Code and Copilot HTTP hooks fail open by vendor design, and Cursor and Codex use a fail-closed command wrapper. For custom agents, the enforcement point is a guard() call in your code, next to the tool. Nothing routes through a proxy, so there's no gateway to size, place, or bypass.

Policies engineers can read and test

Onyx and Arcjet both create policies from natural language, and they differ in the output. Onyx turns a description into enforceable rules aligned to compliance standards. Arcjet generates Rego or builder rules that you review, test, and run in dry run before they go live. An Arcjet policy is an artifact that engineers can read in a code review and that behaves the same way every time it evaluates.

Public documentation and self-serve start

Arcjet's documentation, pricing, and SDKs are public, with a 15-day trial and published plans that start at $25 per month. An engineer can install the hooks or the SDK and see decisions the same day. Onyx's documentation requires a sign-in and its pricing isn't published, which fits an enterprise sales motion.

Where Onyx goes further

Onyx covers areas that Arcjet doesn't. Onyx lists integrations with SaaS agent platforms such as Microsoft Copilot Studio and Salesforce, where Arcjet can't enforce. Onyx also describes discovery across SaaS, cloud, endpoints, and code, browser AI coverage, posture management, automated red teaming, model routing, and ROI reporting.

When to choose Onyx Security

Choose Onyx Security when you need one enterprise governance program across all employee AI use, bought through a sales process. Onyx fits when any of the following apply:

• You need one governance program across employee AI use, including browser AI and SaaS agent platforms such as Microsoft Copilot Studio.
• Enterprise-wide discovery of sanctioned and shadow AI is your first requirement.
• You want posture management and automated red teaming from the same vendor.
• You want an AI gateway for model routing and cost optimization alongside security.
• You buy security through an enterprise sales process and expect a guided deployment.

When to choose Arcjet

Choose Arcjet when your engineers own the agent and you need a documented, testable decision before each action. Arcjet fits when any of the following apply:

• You want to enforce one policy on Claude Code, GitHub Copilot, Cursor, and OpenAI Codex before each tool call runs, with no proxy to deploy.
• You build custom agents in JavaScript, TypeScript, Python, or Go and want the decision in your code, next to the tool.
• You want policies in Rego with tests and dry run, published without an application deploy.
• You want to check the destination an agent is about to contact against threat intelligence.
• You also need bot detection, rate limiting, and WAF rules for the application that hosts the agent.
• You want to read the documentation and start without a sales call.

For more information about how the layers of agent security fit together, see AI agent security architecture.

Can you use Onyx Security and Arcjet together?

Yes, Onyx Security and Arcjet can run side by side. Onyx can provide the enterprise inventory, posture, red teaming, and governance program across SaaS and browser AI. Arcjet can provide the enforcement decision in the places where your engineers own the agent: coding agents on developer machines and the tool calls inside your own agents. Arcjet decisions can reach the same SIEM as Onyx through export to Datadog, Splunk, SentinelOne, Panther, or Amazon S3 on the Enterprise plan.

What are the alternatives to Onyx Security?

Onyx Security alternatives include other enterprise AI security platforms and tools that enforce policy on agent actions directly. The following products are common alternatives to Onyx Security, depending on which part of Onyx you need:

• Arcjet enforces one policy on coding agents through their native hooks and on custom agents through SDKs, with public documentation and self-serve pricing.
• Runlayer is an AI enablement and control platform built around an MCP gateway, endpoint discovery, and hooks.
• Pillar Security covers AI discovery, posture management, red teaming, and runtime guardrails.
• Noma Security discovers endpoint agents such as Claude Code, Cursor, and Codex and the MCP servers they use.

For the difference between posture products and runtime enforcement, see AI security posture management versus runtime security.