{
 "license": "CC-BY-4.0",
 "attribution": "Fide AI, Agentic Cyber Explorer",
 "url": "https://agentic-cyber-explorer.pages.dev/events/deltacert-agent-selective-recertification-2026/",
 "asOf": "2026-09-26",
 "id": "deltacert-agent-selective-recertification-2026",
 "date": "2026-08-12",
 "datePrecision": "day",
 "title": "DeltaCert-Agent proposes selective security retesting of LLM agents after configuration changes",
 "lane": "defense",
 "kind": "paper",
 "summary": "An author project page describes DeltaCert-Agent, which maps configuration changes in tool-using LLM agents to affected security claims and reruns only scoped tests plus sentinel checks, escalating to full recertification when impact cannot be bounded. The author reports 75.02% regression-detection recall versus 55.01% for equal-budget random selection while running 61.35% fewer tests, using four small locally hosted models.",
 "whyItMatters": "Continuous agent changes make full security re-evaluation costly, and this work tests a cheaper recertification strategy.",
 "actors": [],
 "topics": [
  "eval-validity",
  "tool-and-mcp-security"
 ],
 "atlas": [
  "tools",
  "eval-environment"
 ],
 "artifacts": [
  "llama"
 ],
 "sources": [
  {
   "url": "https://dranubhaparashar.github.io/projects/posts/deltacert_agent/deltacert_agent/",
   "publisher": "Author project page (Anubha Parashar)",
   "title": "DeltaCert-Agent",
   "date": "2026-08-12",
   "type": "primary",
   "accessed": "2026-09-25"
  }
 ],
 "keyFacts": [
  {
   "fact": "75.02% regression-detection recall vs 55.01% for equal-budget random selection; 61.35% fewer tests executed on average.",
   "locator": "Project page, results"
  },
  {
   "fact": "31,396 evidence rows across Qwen3, Gemma3, Llama 3.2 and Phi-4 Mini over five repetitions.",
   "locator": "Project page, evaluation setup"
  }
 ],
 "significance": 1,
 "fideQuestions": [
  "FID-075"
 ],
 "methods": [
  "sandboxing-egress"
 ],
 "review": "assistant-drafted",
 "addedOn": "2026-09-25"
}