Arcjet vs Noma Security

How do Arcjet and Noma Security differ?

Noma Security is an enterprise AI and agent security platform that combines posture management, agent access control, detection and response, and red teaming, and enforces through agent hooks, gateways, SDKs, and APIs. Arcjet is an AI agent runtime security platform that enforces policy in the path of each action: in your code for custom agents, and in the hooks of Claude Code, GitHub Copilot, Cursor, and OpenAI Codex, with no proxy to route traffic through. Both products enforce on coding agents through hooks, and the key difference is scope. Choose Arcjet if you want policy as code in Rego, enforcement at the tool call in your own agents, web application security in the same SDK, and published pricing. Choose Noma if you want posture management, red teaming, and coverage of software as a service (SaaS) agents such as Copilot Studio and Agentforce from one enterprise vendor.

Arcjet publishes this comparison. Competitor details come from Noma Security's public documentation, reviewed on September 25, 2026. Products change, so check the linked sources before you decide.

Arcjet and Noma Security at a glance

Arcjet and Noma Security both enforce through coding-agent hooks, and Noma adds posture management, red teaming, and SaaS agent coverage. The following table compares the two products across 12 areas.

What is Noma Security?

Noma Security is an enterprise AI and agent security platform built from four modules. Noma's home page lists them as AI security posture management (AI-SPM), which discovers agents, MCP servers, skills, and tools; Agent Access Control; AI detection and response (AI-DR), which monitors "the full behavioral chain of every agent session"; and AI red teaming with multi-turn attacks. Noma calls its enforcement model Open Enforcement: you define governance once and Noma enforces it "through agent hooks, MCP gateways, AI gateways, agent SDKs, and direct APIs."

For coding assistants, Noma's coding-assistant page names five agents: Claude Code, Cursor, GitHub Copilot, Codex, and Windsurf. The page describes enforcement through native hooks before and after tool calls and shell commands, deployment through Cursor Enterprise or MDM tools such as Jamf and Intune, an integration with CrowdStrike EDR, and masking of detected sensitive data. The same page says Noma evaluates every prompt, tool call, and model output.

Noma publishes its coding-agent plugins on GitHub under the Apache 2.0 license. The plugins stream hook events to the Noma backend, which returns allow, block, or flag decisions. Noma's endpoint agents page describes discovery through your existing EDR or MDM, including agents, MCP servers, skills, and connected personal accounts.

Beyond coding agents, Noma's agent control plane page names Copilot Studio, Agentforce, Bedrock, Azure AI Foundry, and Databricks, and more than 80 connectors. On June 2, 2026, Noma launched Agentic Access Control, an agent registry with Approved, Requires Review, and Blocked states and tool-level control. Noma's Claude ecosystem article, published July 15, 2026, describes coverage of Claude Chat through the Claude Compliance API and of Claude Cowork through OpenTelemetry.

Noma raised a $100 million Series B led by Evolution Equity Partners, announced July 31, 2025. Noma's website doesn't publish pricing and offers a demo request.

What is Arcjet?

Arcjet is an AI agent runtime security platform. Arcjet discovers the agents running in your organization, enforces policy on every action, prompt, and tool call, and keeps the evidence of what happened. Arcjet decides in the path of the action rather than on the network.

For coding agents, Arcjet enforces policy on Claude Code, GitHub Copilot, Cursor, and OpenAI Codex through the hooks each agent already fires, with no SDK and no code change. One policy applies across all four agents, and you choose when it runs: Tool call, Prompt, or Model switch. You install the hooks through managed settings or MDM, so developers can't remove them without administrator access.

Arcjet's 12 starter policies cover destructive commands, history rewrites, protected paths, credential access, piped installers, MCP allowlists, egress allowlists, destination threats, model allowlists, sensitive information, prompt injection, and npm publish. Evaluation runs at the edge in over 300 data centers, and a hook typically adds a few tens of milliseconds. Arcjet also records sessions without a code change from OpenTelemetry and the Claude Compliance API.

For custom agents, the JavaScript, TypeScript, Python, and Go SDKs call guard() on tool calls, queue consumers, MCP handlers, and other actions that have no HTTP request, and protect() on HTTP routes. Arcjet integrates with 15 agent frameworks, including the Claude Agent SDK, CrewAI, Google ADK, LangChain, LangGraph, Mastra, OpenAI Agents, Strands Agents, the Vercel AI SDK, and Microsoft Agent Framework for Go.

You write Guard policies in Rego or a builder, or generate one from an English description, dry run it before it goes live, and publish it from the Arcjet Console or MCP server. Changes take effect in real time.

Arcjet detectors include prompt injection, sensitive information with an SDK-local option that keeps the raw body in your process, destination threat analysis, token budgets, and rate limits. The same SDK covers bot detection, Shield WAF, and email validation for your web application. The Arcjet Console records every decision and session, and SIEM export to Datadog, Splunk, SentinelOne, Panther, and Amazon S3 is available on the Enterprise plan.

Key differences between Arcjet and Noma Security

Arcjet differs from Noma Security in four ways: enforcement inside your own agents, policy as code, the handling of coding agents and tool results, and coverage of web security, data handling, and pricing. The following sections cover each one.

Enforcement inside your own agents

Noma lists agent SDKs and direct APIs among its enforcement points, alongside gateways. Arcjet's model for custom agents is a check at the tool call, in the same function that performs the side effect, so a policy can read the tool arguments, the user, the account, and the plan.

The Arcjet check needs no gateway, and it covers work with no model traffic behind it, such as a queue consumer or a scheduled job. Arcjet documents integrations for 15 agent frameworks in JavaScript, TypeScript, Python, and Go. For more information about this design, see AI agent security architecture.

Policy as code

Arcjet policies are Rego, the Open Policy Agent language, or a builder that generates Rego. Each Arcjet policy is reviewable text with tests, dry run before it can deny, and a publish step that takes effect in real time. A denial on a coding agent names the rule ID, the policy, and the decision, so the developer who was blocked and the reviewer can trace it to the same record. Noma's public documentation describes a central control plane where you define governance once, and doesn't describe a policy language.

Coding agents and tool results

Both products enforce through the hooks of Claude Code, Cursor, GitHub Copilot, and Codex, and Noma also names Windsurf. Noma's coding-assistant page says Noma evaluates model output as well as prompts and tool calls. Arcjet screens prompts and tool calls and records tool results as activity, but Arcjet doesn't screen tool results before they reach the model on coding agents.

Arcjet documents the vendor limits of each hook, including that Claude Code and Copilot HTTP hooks fail open by vendor design and that Arcjet uses a fail-closed command wrapper for Cursor and Codex. For more information, see how to secure AI coding agents.

Web security, data handling, and pricing

Arcjet also covers bots, rate limiting, and Shield WAF for the web application that hosts your agent, which Noma's public documentation doesn't describe. Arcjet publishes its prices, from $25 per month, and offers a 15-day trial.

Arcjet's SDK-local sensitive-information detection keeps the raw body in your process. Prompt injection detection runs in the Arcjet Cloud API, so the local-processing claim applies to sensitive-information inspection, not to every rule. For more information, see keeping security inspection local.

When to choose Noma Security

Choose Noma Security when you want posture, red teaming, detection, and SaaS agent coverage from one enterprise vendor. Noma fits when any of the following apply:

• You want posture management, red teaming, and runtime detection from one enterprise vendor.
• You need coverage of SaaS agent platforms such as Copilot Studio and Agentforce.
• You want endpoint discovery that reuses your existing EDR or MDM, including detection of personal accounts.
• Evaluating model output and tool results is a firm requirement.
• You run Windsurf alongside other coding assistants.

When to choose Arcjet

Choose Arcjet when you want policy as code and enforcement at the tool call, in your own agents and on coding agents. Arcjet fits when any of the following apply:

• You build your own agents and want enforcement at the tool call, with application context, and no gateway.
• You want policy as code in Rego, with tests and dry run before live.
• You want one policy that decides before the tool call runs on Claude Code, GitHub Copilot, Cursor, and Codex.
• You want bots, rate limiting, and WAF for the same application.
• You want published, usage-based pricing and a self-serve start.

Can you use Noma Security and Arcjet together?

Yes, Noma Security and Arcjet can run side by side. A common split is Noma for posture, red teaming, and SaaS agent platforms, and Arcjet for in-code enforcement in your own agents and web application. Both products can install coding-agent hooks, so choose one owner for each hook event on each agent rather than running two policies on the same event.

What are the alternatives to Noma Security?

Noma Security alternatives include other enterprise AI security platforms, platform-native controls, and runtime enforcement in code. The following products cover overlapping ground:

• Arcjet provides in-path enforcement for custom agents in code and for Claude Code, GitHub Copilot, Cursor, and Codex through their hooks, plus web application security in the same SDK.
• Lasso Security provides discovery, posture, red teaming, and runtime enforcement at a proxy, API, or AI gateway layer, plus coding-agent hooks. For more information, see Arcjet vs Lasso Security.
• Microsoft Agent 365 is a control plane that extends Defender, Entra, and Purview to agents, suited to Microsoft 365 organizations. For more information, see Arcjet vs Microsoft Agent 365.
• Zenity provides observability, posture management, runtime enforcement, and threat detection for agents across SaaS, cloud, and endpoint deployments. For more information, see Arcjet vs Zenity.

For a wider map of the category, see AI agent security platforms.