mlsanityData to 14 Oct 2026

PocketAgents: A Manifest-Driven Library of Autonomous Defense Agents

arXiv 2605.21694 ↗First posted 20 May 2026cs.CR cs.AIcited by 0

On the map: LLM agents and tool use → Agent threats, guardrails and governance

Abstract (arXiv)

Connecting large language models (LLMs) to defensive enforcement requires more than asking a model whether an attack is happening. A defender must decide which model outputs may change the system state, which outputs must be rejected, and how failures should be recorded. We present PocketAgents, a manifest-driven library of autonomous defense agents. Each agent is installed as three data files: a manifest, a prompt, and a runtime context. The shared runtime gives the agent bounded telemetry access and accepts only typed reports whose requested action appears in the manifest. We implemented PocketAgents on top of Perry, a cyber-deception testbed, and evaluated two agents for the Command and Control and Exfiltration tactics in 18 closed-loop trials of a DarkSide-inspired attack on a small enterprise topology. Thirteen trials produced validated network-block actions and contained the attack; four failed schema validation; one produced a valid no-action decision. The experiments show that a typed boundary makes LLM-driven defense measurable, extensible, and attributable.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use. Builds on 13 of its 30 references that are in the map. First posted on arXiv 20 May 2026.

Builds on

13 of 30 references.

Closest ancestors · 7

Same direction, newest first

  1. 2026Agent-Sentry: Bounding LLM Agents via Execution Provenance24 Mar
  2. 2026AIR: Improving Agent Safety through Incident Response12 Feb
  3. 2025Progent: Securing AI Agents with Privilege Control16 Apr
  4. 2025AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agents24 Mar
  5. 2024Agent-SafetyBench: Evaluating the Safety of LLM Agents19 Dec
  6. 2024When LLMs Go Online: The Emerging Threat of Web-Enabled LLMs18 Oct
  7. 2024IsolateGPT: An Execution Isolation Architecture for LLM-Based Agentic Systems8 Mar

Other references · 6

  1. 2024AutoAttacker: A Large Language Model Guided System to Implement Automatic Cyber-attackscited by 96
  2. 2026Towards a Science of AI Agent Reliabilitycited by 65
  3. 2020UNICORN: Runtime Provenance-Based Detector for Advanced Persistent Threatscited by 53
  4. 2024CTINexus: Automatic Cyber Threat Intelligence Knowledge Graph Construction Using Large Language Modelscited by 14
  5. 2025Perry: A High-level Framework for Accelerating Cyber Deception Experimentationcited by 1
  6. 2026SOCpilot: Verifying Policy Compliance for LLM-Assisted Incident Responsecited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Autonomous LLM Agents & CTFs: A Second Look
  2. 2026The Landscape of Prompt Injection Threats in LLM Agents: From Taxonomy to Analysis
  3. 2026Beyond Handcrafted Security: Towards Self-Evolving Defense for LLM Agents
  4. 2026Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation
  5. 2026Autonomous Adversary: Red-Teaming in the age of LLM
  6. 2025The Dark Side of LLMs: Agent-based Attack Vectors for System-level Compromise
  7. 2026BlueSTAR: Tiered Agentic Architecture for Autonomous Cyber Defense
  8. 2026Systematic Capability Benchmarking of Frontier Large Language Models for Offensive Cyber Tasks