Arcjet vs Prompt Security

How do Arcjet and Prompt Security differ?

Prompt Security is SentinelOne's AI security product. Prompt Security governs how employees, developers, and applications use AI tools, with browser extensions, a lightweight agent, an MCP gateway, and an API for homegrown apps that redact sensitive data and block risky prompts. Arcjet is an AI agent runtime security platform that enforces policy in the path of each action: inside your code for the agents you build, and in the hooks that Claude Code, GitHub Copilot, Cursor, and OpenAI Codex already fire. The key difference is data versus actions. Prompt Security controls what data reaches an AI tool, and Arcjet decides whether an agent can take an action. Choose Prompt Security when your main concern is what employees paste into the over 15,000 AI tools it covers. Choose Arcjet when you need a policy on what agents do, such as the commands, file reads, destinations, and MCP calls of coding agents and your own agents, with no proxy to route through.

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

Prompt Security and Arcjet at a glance

Prompt Security inspects what employees and applications send to AI tools, and Arcjet decides what agents do in your code and in coding-agent hooks. The following table compares the two products across 12 areas.

What is Prompt Security?

Prompt Security is an AI security product, owned by SentinelOne, that governs employee, developer, and application use of AI tools. SentinelOne announced its agreement to acquire Prompt Security on August 5, 2025, and its quarterly filing records completion on September 5, 2025, for about $133.6 million in cash plus 1,555,099 Class A shares and 415,109 assumed options.

Itamar Golan and Lior Drihem founded Prompt Security in 2023, and the company raised an $18 million Series A led by Jump Capital in November 2024. The product keeps the Prompt Security name on the SentinelOne product page, and prompt.security still serves research and open source tools such as Prompt Fuzzer.

At OneCon on November 5, 2025, SentinelOne launched four products: Prompt Security for Employees, for AI Code Assistants, and for AI Applications as generally available, and Prompt Security for Agentic AI in beta. On March 23, 2026, SentinelOne announced Prompt AI Agent Security in preview and Prompt AI Red Teaming.

The acquisition announcement describes three delivery mechanisms: a lightweight agent, browser extensions, and an MCP gateway, across browsers, desktop applications, and APIs. The product page claims coverage of over 15,000 AI tools and services, and Prompt Security can block, redact, or warn when a prompt violates policy.

For code assistants, Prompt Security redacts and sanitizes code to keep secrets, PII, and intellectual property out of assistants such as GitHub Copilot and Cursor, and the SentinelOne page also names Claude Code. The MCP gateway sits between AI applications and more than 13,000 known MCP servers, assigns each a risk score, and allows or blocks by user, server, or action. For homegrown apps, Prompt Security inspects prompts and responses for prompt injection, jailbreaks, and data leaks, deployed in the cloud, in your VPC, or on premises.

Prompt Security is sold as part of the SentinelOne Singularity Platform. SentinelOne doesn't publish a price list; its AWS Marketplace listing prices per user and directs buyers to contact SentinelOne.

What is Arcjet?

Arcjet is an AI agent runtime security platform. Arcjet discovers the agents that run in your organization, enforces policy across every action, prompt, and tool call, and keeps the evidence of 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 enforces policy on Claude Code, GitHub Copilot (CLI and cloud agent), 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 decides before the tool call runs. You choose when the policy runs with Execute on: Tool call, Prompt, or Model switch.

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. You install the hooks through managed settings or mobile device management (MDM), so developers can't remove them without administrator access. For visibility without a hook, Arcjet ingests OpenTelemetry and the Claude Compliance API.

For agents and applications you build, Arcjet Guards run inside your code. The JavaScript, TypeScript, Python, and Go SDKs call guard() on tool calls, queue consumers, MCP handlers, and any action with no HTTP request, and protect() on 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, the Vercel AI SDK, Vercel Eve, and Microsoft Agent Framework for Go.

Arcjet policies are managed centrally, written in Rego or with a builder, and published from the Arcjet Console or the Arcjet MCP server. Each rule runs in dry run before it goes live, and changes take effect in real time. You can also generate a policy from an English description.

Arcjet detectors include prompt injection, sensitive information, destination threat analysis, token budgets, and rate limits. The same SDK covers bot detection, Shield WAF, rate limiting, and email validation for web applications.

Hooks have limits that the agent vendors set. Claude Code and Copilot HTTP hooks fail open by vendor design, and Cursor and Codex use a fail-closed command wrapper. Personal accounts don't load organization hooks, so you block those at the network or device. For details, see why coding agent hooks fail open.

Key differences between Arcjet and Prompt Security

Arcjet differs from Prompt Security in four ways: whether each product controls data or actions, how each handles coding agents, where enforcement sits, and how each is bought and deployed. The following sections cover each one.

Data leaving versus actions taken

Prompt Security's core question is what data reaches an AI tool. Prompt Security's browser extension and agent inspect prompts to over 15,000 AI tools and services and redact what shouldn't leave.

Arcjet's core question is whether an agent can take an action. A coding agent that runs rm -rf, reads ~/.aws/credentials, pipes an installer to a shell, or calls an unapproved MCP server is taking an action, and the Arcjet policy decides before the tool call runs. For examples, see how to block dangerous commands in coding agents and how to stop coding agents reading secrets.

Coding agents

Prompt Security for AI Code Assistants focuses on redacting secrets, PII, and intellectual property from code, and on scanning AI-generated output for vulnerable code. Prompt Security's public documentation doesn't describe a policy engine over coding-agent tool calls, such as shell commands, file paths, or network destinations. Arcjet provides that policy through the agents' own hooks, installed through managed settings, with one policy across Claude Code, Copilot, Cursor, and Codex. For a full walkthrough, see how to secure AI coding agents.

Where enforcement sits

Prompt Security uses endpoint and browser sensors and gateways that sit between the user or app and the AI service. Arcjet has no proxy, browser extension, or endpoint agent. For agents you build, the Arcjet check is a function call in your tool handler, so the check runs with your application's context, such as the user, the tenant, or the resource, and it covers tool calls that never cross a gateway. For the trade-off, see PII detection at the gateway versus in the application.

Buying and deployment

Prompt Security is purchased through SentinelOne, priced per user, with self-hosted and on-premises options. Arcjet has self-serve plans (Individual at $25 per month, Startup at $299 per month) and custom Enterprise pricing, with usage at $50 per million agent requests. For more information, see Arcjet pricing.

When to choose Prompt Security

Choose Prompt Security when your main risk is sensitive data that employees send to AI tools. Prompt Security fits when any of the following apply:

• Your main risk is employees pasting sensitive data into ChatGPT, Gemini, and other AI tools, and you want browser-level redaction and coaching.
• You want to discover shadow AI usage across a workforce.
• You already run SentinelOne Singularity and want AI usage controls in the same platform.
• You need a self-hosted or on-premises deployment for regulated or air-gapped environments.

When to choose Arcjet

Choose Arcjet when your main risk is what agents do: the commands, files, destinations, and MCP calls of coding agents and your own agents. Arcjet fits when any of the following apply:

• You need to decide which commands, files, destinations, MCP servers, and models your developers' coding agents can use.
• You build agents in code and want the check to run inside the tool handler, with your application's context.
• You want to score the destinations an agent is about to contact with threat intelligence.
• You want Rego policies with tests and dry run, published from the Arcjet Console, CLI, or MCP server.
• You also need bot detection, rate limiting, and a WAF for the web application that hosts the agent.
• You want self-serve pricing and a free plan.

Can you use Prompt Security and Arcjet together?

Yes, Prompt Security and Arcjet cover different layers and can run together. Prompt Security can govern what employees send to AI tools in the browser and on the desktop, and Arcjet enforces policy on what coding agents and your own agents do. Arcjet exports decisions to SentinelOne, Datadog, Splunk, Panther, and Amazon S3 on the Enterprise plan, so Arcjet decisions can sit next to Prompt Security and SentinelOne telemetry for detection and alerting.

What are the alternatives to Prompt Security?

Prompt Security alternatives include agent governance platforms, broad AI security platforms, hosted detectors, and runtime enforcement in code. The following products are common alternatives to Prompt Security, depending on which control you need:

• Arcjet provides runtime enforcement in your code and in the hooks of Claude Code, GitHub Copilot, Cursor, and Codex, with self-serve pricing.
• Zenity provides agent inventory, posture, and runtime boundaries, and is strongest on low-code and software as a service (SaaS) agents such as Copilot Studio.
• Palo Alto Networks Prisma AIRS is a broad AI security platform with runtime scanning, an AI gateway, model security, and red teaming.
• Lakera, part of Check Point, is a hosted detector for prompt injection and data leakage that you call from your application.

For a broader shortlist, see AI agent security platforms.