Blog

All articles

Every deep-dive, teardown and field note from the lab — newest first.

  1. Secure Agentic Engineering: Security by Design for AI Agents

    Why agent security has to be designed alongside the architecture, not bolted on afterward — the concrete engineering-process points where the outcome of every future attack is actually decided.

    Read
  2. Agent Tool Sandboxing: seccomp, gVisor, WASM, and microVMs as Layers of Containment

    How to sandbox AI agent tools with seccomp, gVisor, WASM, and microVMs. A layered containment model that isolates untrusted-data processing and limits blast radius.

    Read
  3. AI Agent Guardrails: Input and Output Filtering Against Prompt Injection

    How AI agent guardrails work: input/output validation, prompt injection detection, content policies, and pre-action output checks that stop an agent before it acts.

    Read
  4. AI Agent Incident Teardowns: Anatomy of Real Prompt-Injection and Exfiltration Breaches

    Three real AI agent incidents torn apart: EchoLeak, Slack AI, and GitHub MCP. The attack chains, the root causes, and the lessons learned for defenders.

    Read
  5. AI Agent Security Benchmarks: Mapping Your Eval Harness to MITRE ATLAS and the OWASP Top 10 for Agentic Applications

    How to map AI agent security benchmarks and your eval harness to MITRE ATLAS techniques and the OWASP Top 10 for Agentic Applications (ASI). A practical guide.

    Read
  6. The AI Agent Threat Landscape in 2026: Attack Surface, Emerging TTPs, and What Changed This Year

    A field guide to the 2026 AI agent threat landscape: the autonomous-agent attack surface, the emerging TTPs that defined this year, and what actually changed.

    Read
  7. Capability-Based Security for Agent Tools: Enforcing Least Privilege with Scoped, Per-Tool Tokens and Object Capabilities

    How capability-based security, object capabilities, and scoped per-tool tokens let you enforce least privilege for AI agents, so a compromised tool cannot reach what it was never handed.

    Read
  8. Cryptographic Identity for AI Agents: Issuing Verifiable Identity with mTLS, JWT, and SPIFFE/SPIRE

    How to give non-human AI agents verifiable cryptographic identity using mTLS, JWT, and SPIFFE/SPIRE — why static API keys fail and how to issue short-lived workload identity.

    Read
  9. CTF-Style Challenges for AI Agent Security: Building Reproducible Playgrounds and Scoring Agent Exploitation

    How to design AI security CTF challenges that are reproducible, score real agent exploitation, and go beyond prompt injection puzzles to test tool-using agents.

    Read
  10. Defense-in-Depth for Autonomous AI Agents: A Complete Layered Security Architecture

    A complete guide to defense-in-depth for autonomous AI agents: guardrails, I/O filtering, least-privilege tools, sandboxing, and anomaly monitoring in one model.

    Read
  11. Delegation Chains and the Confused Deputy in Agent Identity: Safe Impersonation, Token Exchange, and Audit Trails

    How AI agents pass authority safely down a delegation chain, why the confused deputy bites, and how token exchange (RFC 8693) keeps audit trails intact.

    Read
  12. Industry-Specific AI Agent Risk: How the Threat Model Shifts for Finance, Healthcare and Legal

    Industry AI agent risk is not generic. See how the threat model shifts for finance agent security, healthcare AI agents and legal AI agent deployments.

    Read
  13. Least-Privilege Agent Tools: Capability Scoping, Permission Gating, and Human-in-the-Loop

    How to give AI agents only the tools they need: capability scoping, permission gating, dynamic toolset narrowing, and human-in-the-loop checkpoints for irreversible actions.

    Read
  14. Malicious MCP Servers: How Rogue Tool Servers Cross Your Agent's Trust Boundary

    A malicious MCP server runs with your agent's privileges and dictates its tool behavior. How rogue servers get in, what they can do, and how to verify, pin, and isolate them.

    Read
  15. Mapping AI Agent Attacks to MITRE ATLAS: Tactics, Techniques, and the Detection Signals Each Phase Produces

    A practical walkthrough of mapping AI agent attacks to MITRE ATLAS tactics and techniques, plus the concrete detection signals each phase leaves behind for your SOC.

    Read
  16. MCP Security: Tool Poisoning, Rogue Servers, and Supply-Chain Exploits

    How attackers exploit the Model Context Protocol: tool poisoning, malicious and rogue MCP servers, rug pulls and supply-chain attacks against agent tooling — the offensive threat model.

    Read
  17. MCP Server Isolation: Containing Untrusted Tool Servers and the CVEs That Prove Why

    Why MCP servers are the new untrusted dependency: a tour of the CVE classes — command injection, tool poisoning, path traversal, SSRF — and how to isolate, scope, and sandbox each server so a compromised tool can't own your agent.

    Read
  18. Testing Memory Poisoning in a Red-Team Lab: Seeding RAG Corpora and Detecting Tainted Retrieval

    Build a memory poisoning test in your red-team lab: seed RAG corpora and vector stores with tainted records, then detect poisoned retrieval before it reaches the model.

    Read
  19. Novel Attack Classes for Autonomous Agents: Consensus Hijacking and Context Boundary Confusion

    A field guide to novel agent attacks — multi-agent consensus hijacking and context boundary confusion — plus the emerging AI threats security teams should track now.

    Read
  20. OAuth 2.1 and OIDC for AI Agents: Client Credentials, Token Rotation, and Tight Scopes

    How to authenticate AI agents with OAuth 2.1 and OIDC: when to use the client-credentials flow, how token rotation limits blast radius, and how to design scopes that stay tight.

    Read
  21. The OWASP Top 10 for Agentic Applications (ASI01–ASI10): Attack Vectors and Exploitability of Each Risk

    An attacker's-eye analysis of the OWASP Top 10 for Agentic Applications (ASI01–ASI10): how each risk is actually exploited, the attack vector behind it, and what makes an agent vulnerable.

    Read
  22. Red-Teaming Prompt Injection Against AI Agents: Payloads, Evasion, and TTPs in the Lab

    A hands-on lab playbook for red-teaming prompt injection against AI agents: building jailbreak payloads, indirect injection testing, evasion patterns, and attacker TTPs.

    Read
  23. Offensive Red-Team Frameworks for AI Agents: Garak, PyRIT, and Promptfoo in the Lab

    A hands-on offensive comparison of Garak, PyRIT, and Promptfoo for red-teaming AI agents in the lab: what attack techniques each tool automates, how they differ, and when to reach for which.

    Read
  24. Token Theft and Replay Attacks Against AI Agents: How Agent Credentials Leak and How to Contain the Blast Radius

    How AI agent credentials leak, why stolen tokens replay past MFA, and how token binding plus short-lived tokens shrink the blast radius of token theft.

    Read
  25. Tool-Call Fuzzing and Abuse Testing for AI Agents: Finding Excessive Agency and Confused-Deputy Paths

    A red-team guide to tool call fuzzing and function call abuse testing for AI agents: how to surface excessive agency and confused-deputy paths before attackers do.

    Read
  26. Tool Description Poisoning in MCP: How Rug Pulls and Hidden Instructions Hijack Your Agent

    How MCP tool description poisoning and rug pulls hide malicious instructions in tool metadata, plus pinning and audit defenses that actually catch post-install mutation.

    Read
  27. Zero-Trust for AI Agents: the Trust Assumptions Attackers Exploit — and the Enforcement That Stops Them

    Why implicit trust between AI agents is an attack path: how adversaries abuse agent-to-agent trust and delegated identity, and how continuous verification, a policy decision point and mTLS shut it down.

    Read
  28. The AI Agent Attack Surface: A Practical Threat Model

    A practical threat model of autonomous AI agent security: prompt injection, tool abuse, memory poisoning, excessive agency, and multi-agent propagation — mapped to the OWASP Top 10 for Agentic Applications and MITRE ATLAS.

    Read
  29. Indirect Prompt Injection in AI Agents: Attack Chains, Payloads, and Injection Techniques

    How attackers chain indirect prompt injection to hijack AI agents: injection sources, payload construction, multi-step attack chains, and the techniques that carry hidden instructions into the model.

    Read
  30. Memory and Context Poisoning in AI Agents: Attack Patterns and Injection Techniques

    How attackers poison an AI agent's long-term memory and retrieved context: injection patterns, persistence techniques, and how false data is planted to hijack every future session.

    Read
  31. Multi-Agent and Agent-to-Agent Attacks: How One Compromised Agent Poisons the Swarm

    How a single compromised agent spreads malicious instructions through a multi-agent system, which trust assumptions enable lateral movement, and what isolation defenses can contain the blast radius.

    Read
  32. Tool and Function-Call Abuse in AI Agents: The Confused-Deputy Problem Explained

    How attackers exploit AI agents' privileged tool access through the confused-deputy problem — mechanism, concrete example, and least-privilege defenses that actually work.

    Read