// Archive
Agentic AI security
Security engineering for AI agents: prompt injection, agent memory, MCP, tool-calling, workload identity, and agent architecture attack surface.
Categories
- NOT-014AI Agent Memory Poisoning: How to Audit and Harden Your AgentOne bad line written to an agent's memory outlives every session. Here's how the write happens, how to audit for it, and how to make the store hostile to poison.#ai safety#agent memory#memory poisoning9 min read
- NOT-013Workload Identity Federation for AI Agents: Kill the Static KeyAgents are the new source of leaked secrets — AI-service keys grew 81% in a year. Workload identity federation replaces the static key with a short-lived token minted from the identity your agent already has. Here's the five-step migration.#workload-identity#agentic-ai#oidc8 min read
- NOT-012Validate Agent Tool Calls Before They ExecuteAgents generate tool arguments by sampling tokens — malformed calls aren't an edge case, they're a statistical certainty. Here's the five-minute contract check that runs before the side effect, not after the incident.#json-schema#agentic-ai#tool-calling5 min read
- NOT-011Open Knowledge Format (OKF): Google's Answer to the AI Agent Context ProblemGoogle Cloud's OKF is a directory of markdown files with YAML frontmatter — nothing more, and that's the point. A format, not another service, so any agent can consume any knowledge base without an SDK. And v0.2 added the trust layer that makes agent-written knowledge safe to act on.#agentic ai#open knowledge format#knowledge management13 min read
- NOT-010Your Multi-Agent Graph Has No BoundariesThe failure mode nobody designs for: one compromised agent acting as every other agent in the graph. A four-boundary checklist for multi-agent deployments.#agentic-ai#security#multi-agent4 min read
- NOT-009Your AI Agent Config Is Leaking SecretsThe files that configure your coding agent — .mcp.json, .claude/settings.json, the system prompt — are plaintext on disk and read by code that executes on your behalf. Most of them contain secrets, over-broad permissions, or both. Here's the five-minute audit.#ai safety#mcp#agent security5 min read
- NOT-008How to Probe an MCP Server (Without Installing Anything)A five-minute, no-install workflow for checking what an MCP endpoint reveals about itself — server identity, exposed tools, auth requirements — before you wire it into an agent.#mcp#agentic-ai#security4 min read
- NOT-007Graph Engineering: The Agent Architecture Shift That Changes Your Attack SurfaceGraph engineering went from a hashtag to a default architecture in a month. The security community is only starting to map what it breaks: shared state, agent-to-agent injection, and trust boundaries that no longer sit at the front door.#agentic ai#graph engineering#prompt injection10 min read
- NOT-006AI Agent Memory Is the New Log4jEvery agent framework has a memory store. Almost nobody audits it. The attack that lives there never sends a single malicious instruction.#ai safety#agent memory#memory poisoning6 min read
- NOT-005AI Governance Just Got TeethThe EU started enforcing the AI Act two weeks ago. Colorado rewrote its law. And Gartner says 40% of agentic AI projects will die from inadequate risk controls. Governance is no longer a slide deck.#ai governance#agentic ai#compliance6 min read
- NOT-004A Security Checklist for MCP Servers and System PromptsAn agent's attack surface has two halves: the system prompt that steers it and the tools it's allowed to call. Most teams only ever check one.#ai safety#mcp#prompt injection4 min read
- NOT-003The Best Agent Harness Is the One You Don't NoticeThe harness isn't the interesting part of an agent system. That's exactly why it's the part worth getting right.#ai architecture#agent harness2 min read
- NOT-002Agentic Systems Aren't Apps With Extra StepsGive a model tools and a loop and you haven't built a feature — you've built a distributed system that talks to itself.#ai architecture#agentic systems2 min read
- NOT-001Prompt Injection Doesn't Need Your PromptThe dangerous instruction rarely comes from your user. It comes from the webpage, PDF, or ticket your agent just read.#ai safety#prompt injection3 min read