{
  "topic": "ao-security",
  "title": "Agent ops: security and sandboxing",
  "claim": "Sandbox, prompt-injection design, and layered guardrails.",
  "notes": [
    {
      "slug": "ao-security-attack-taxonomy",
      "title": "A map of attacks on language agents",
      "url": "https://iinventedlight.com/library/notes/ao-security-attack-taxonomy/",
      "status": "published"
    },
    {
      "slug": "ao-security-gemini-cli-sandbox",
      "title": "Sandboxing and trusted folders in Gemini CLI",
      "url": "https://iinventedlight.com/library/notes/ao-security-gemini-cli-sandbox/",
      "status": "published"
    },
    {
      "slug": "ao-security-guardrail-layers",
      "title": "Guardrail layers for tool-using agents",
      "url": "https://iinventedlight.com/library/notes/ao-security-guardrail-layers/",
      "status": "published"
    },
    {
      "slug": "ao-security-prompt-injection-design",
      "title": "Design agents so prompt injection cannot steer them",
      "url": "https://iinventedlight.com/library/notes/ao-security-prompt-injection-design/",
      "status": "published"
    }
  ],
  "artifact_refs": [
    "/library/artifacts/ao-security-sandbox.table.json"
  ],
  "confidence": "medium",
  "as_of": "2026-11-09",
  "license": "Index of CC-BY-4.0 library notes; underlying source licenses remain on each note",
  "schema": "https://iinventedlight.com/library/schemas/topic-packet.schema.json"
}