What Is AI Agent Runtime Security?
AI agent runtime security for unauthorized tool calls, prompt injection, sensitive data, automated abuse, action sequences, and production cost controls.
12 min readArcjet Learning Center
AI agent runtime security and application protection guides
Learn how to secure production AI workflows, enforce agent actions, protect APIs, stop automated abuse, and control resource use inside your application.
128 guides10 topicsUpdated
Start here
Cornerstone guides that set the threat model and control baseline everything else builds on.
API Security Best Practices
API security best practices for authentication, object authorization, input validation, webhooks, abuse controls, monitoring, and continuous testing.
13 min readHow do I secure AI agents in production?
Screen inbound HTTP and channel text, gate tools and MCP before the side effect, and treat observe-only hooks as a diary. Budgets, local PII, and prompt-injection checks sit on those same two boundaries.
4 min readRate-limiting algorithms compared: token bucket, leaky bucket, sliding window, and fixed window
Token bucket, leaky bucket, sliding window, or fixed window? This guide explains how each rate-limiting algorithm works and when to use it in your API.
20 min readAI Security
Identity, architecture, and runtime controls for agents that call tools, access data, and take actions.
AI Agent Security Architecture: In-Code vs Proxy vs Gateway vs Security Agent
Compare in-code, proxy, AI gateway, and security-agent controls by visibility, enforcement, context, latency, failure modes, and threat coverage.
17 min readAI Agent Identity and On-Behalf-Of Authorization
AI agent on-behalf-of authorization: delegated identity, OAuth token exchange, actor claims, credential downscoping, runtime policy, and audit trails.
20 min readHow to prevent prompt injection in LLM applications
A five-layer defense stack: input screening, instruction and data separation, output validation, privilege limiting, and action gating. What each layer catches and what it misses.
14 min readHow-tos
Short, framework-specific recipes for Eve, Mastra, Claude Agent SDK, OpenAI Agents SDK, LangGraph JS, LangChain Python, and the Vercel AI SDK.
How to screen inbound prompts in Claude Agent SDK
Screen inbound prompts on guardHooks({ inbound }) via UserPromptSubmit. A DENY returns { decision: "block" } and Claude Code erases the prompt.
3 min readHow to secure Eve MCP connections
Secure Eve MCP and OpenAPI connections with guardApproval() on the connection's approval field. There is no local execute to wrap.
4 min readHow do I secure MCP tools in LangGraph?
MCP and unwrapped tools execute inside ToolNode. Graph hooks and HITL pauses cannot stop tool.invoke. guardToolNode is the gate. Wrap the node in place; a copy leaves the original unguarded.
5 min readWeb Security
Application-layer guidance for securing APIs, stopping automated abuse, and controlling resource consumption.
- API security6
- Bot protection8
- Rate limiting4
- Runtime security9
- Application & framework security15
- Security concepts4
- Sensitive data & secrets4
What Is API Abuse? How Bots Exploit APIs
API abuse explained: how bots automate valid operations for fraud, scraping, account takeover, and cost exhaustion, and how application controls stop them.
12 min readWeb app security checklist
A production abuse checklist for Next.js, Remix, and Nuxt: bots, identity-aware rate limits, business-logic misuse, early rejection, monitoring, and in-handler controls.
6 min readWhat Is Runtime Application Security?
Runtime application security enforces controls while the application runs, at the point untrusted input meets your code, with context that pre- and post-runtime tools do not have.
10 min readNew & updated
The latest additions and most recently revised guides.
All guides
The complete library, organized by topic.
AI agent security 64 guides
- Best AI security solution for healthcare and regulated industries
- Best AI security solution for fintech
- What's the best AI security provider for enterprises?
- Eve agent security is three jobs
- How to secure a Mastra agent
- Claude Agent SDK security
- Human approval is not a security policy
- canUseTool is not a policy gate
- A sandbox is not a tool policy
- Mastra guardrails vs an action gate
- The lethal trifecta for AI agents
- How do I secure a RAG application?
- Guardian Agents, explained
- Application-Native vs Remote Security Policies: A Technical Guide
- AI Agent Security Architecture: In-Code vs Proxy vs Gateway vs Security Agent
- What Is AI Agent Runtime Security?
- Runtime Security for LLM Applications: Prompt Injection, Data Exfiltration, Unsafe Actions
- Prompt injection protection for LangChain, LlamaIndex, and Vercel AI SDK apps
- Pre-Runtime vs Post-Runtime AI Security Tooling
- Keeping Security Inspection Local So Data Stays in Your Environment
- How to Enforce Runtime Controls on AI Agents Accessing Enterprise Systems
- How to Enforce Token and Spend Budgets on AI Agents
- How to Prevent Data Exfiltration Through AI Agents
- How to Stop AI Agents Accessing Data They Should Not
- How to detect and redact PII in LLM inputs and outputs
- PII detection at runtime: gateway, sidecar, or in-process
- How to stop users sending PII to an LLM
- How to prevent PII leakage from AI agents
- How to redact sensitive data before sending it to OpenAI or Anthropic
- How to prevent LLMs surfacing confidential employee or customer data
- GDPR and CCPA compliance for LLM applications
- How to prevent prompt injection in LLM applications
- How to sanitize user input before passing it to an LLM
- How to prevent a malicious tool call from hijacking your AI agent
- How to defend against indirect prompt injection in agentic workflows
- How AI security platforms detect prompt injection at runtime
- The best tools to detect and block prompt injection in production
- Best AI security tools for developers
- How to Add Human Review and Approval Gates to AI Agent Actions
- Producing Compliance Evidence for AI Agent Activity
- How to Secure an MCP Server or AI Agent Tool Calls
- Anatomy of an Agent Incident: Three Permitted Actions, One Fraud
- How do I add guardrails to an AI agent that calls external APIs?
- Runtime security for LLM applications: prompt injection, data leakage, and output validation
- Which runtime security tools integrate with LangChain or AutoGPT?
- How do I prevent an AI agent from taking irreversible actions?
- How do I implement the OWASP Top 10 for LLM Applications?
- AI security checklist for production LLM apps
- What is AI security?
- What changes for application security when your app calls an LLM?
- Content moderation for AI applications
- Claude Agent SDK security guide
- Claude Managed Agents security guide
- CrewAI security guide
- Genkit security guide
- Google ADK security guide
- LangChain security guide
- LangGraph security guide
- Mastra security guide
- OpenAI Agents security guide
- Strands Agents security guide
- TanStack AI security guide
- Vercel AI SDK security guide
- Vercel Eve security guide
Agentic identity 1 guides
How-tos 13 guides
- How do I secure AI agents in production?
- How to detect prompt injection in an Eve agent
- How to secure Eve MCP connections
- How to secure Mastra MCP tools
- How to screen inbound prompts in Claude Agent SDK
- How to block Bash in Claude Agent SDK
- How do I secure an OpenAI Agents SDK agent?
- How do I screen inbound prompts in OpenAI Agents SDK?
- How do I secure a LangGraph JS agent?
- How do I secure MCP tools in LangGraph?
- How do I secure a LangChain Python agent?
- How do I secure a Vercel AI SDK agent?
- needsApproval and LangGraph interrupt() are not a security policy
API security 6 guides
Bot protection 8 guides
- AI agent bot management
- What Is API Abuse? How Bots Exploit APIs
- How to identify AI agents and bots
- Bot detection techniques for developers
- Is CAPTCHA still effective, and what should you use instead?
- What is bot spoofing and how do you detect it?
- How do you detect bots at the application layer?
- How to protect a React Hook Form from spam
Rate limiting 4 guides
Runtime security 9 guides
- What Is Runtime Application Security?
- SDK-based security vs WAF vs API gateway
- How to Enforce Security Rules at Runtime in Your Code
- How do I detect and block attacks at runtime?
- Are Next.js server actions a security risk?
- Were you affected by the Next.js middleware bypasses?
- How do you prevent SQL injection and XSS in Node.js?
- Does Next.js need a WAF?
- What is permissions-based security in Next.js?
Application & framework security 15 guides
- Next.js security checklist
- Remix security checklist
- Web app security checklist
- How to secure login pages
- Security advice for self-hosting Next.js in Docker
- How to secure your NestJS application with Arcjet
- Structured logging in JSON for Next.js
- Protecting self-hosted Coolify apps
- How do I deploy an Arcjet-protected app to Fly.io?
- How do I secure a container deployment?
- Why do TLS certificates fail in slim Node.js containers on OrbStack?
- How do I detect the client IP on Firebase?
- Why do NEXT_PUBLIC_ variables leak into the Next.js client bundle?
- How do I test Next.js App Router API routes?
- How do I functionally test security rules?
Security concepts 4 guides
Sensitive data & secrets 4 guides
Security in your code
Put these guides into practice with Arcjet
Arcjet brings bot detection, rate limiting, prompt-injection screening, PII detection, and attack protection into your application code, so the controls these guides describe run in the path of the request.