Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Agent Hacks Agent: Autoresearch for Production-Agent Red-Teaming

Source

arXiv

arxiv.org/abs/2607.11698v1

Read the source

Source

Plain-Language Brief

Researchers present AHA, an automated red-teaming approach for production language models like Claude Code and Codex. AHA discovers reusable vulnerability knowledge about other agents by proposing a vulnerability hypothesis, constructing a falsifier, executing the attack in a sandboxed harness, reflecting on the trajectory, and promoting confirmed findings into a Vulnerability Concept Graph (VCG). The frozen VCG outperforms the strongest frozen discovery baseline by 14.2 percentage points and provides an auditable artifact for production safety teams.

What the Document Says

  • The research focuses on automated red-teaming for production language models like Claude Code and Codex.
  • AHA is presented, a loop that proposes a vulnerability hypothesis, constructs a falsifier, instantiates a valid attack, executes it in a sandboxed harness, reflects on the trajectory, and promotes confirmed findings into a Vulnerability Concept Graph (VCG).
  • Each concept in the VCG links an attacker-facing surface to an unsafe trajectory through a claim, enabling condition, falsifier, transfer prediction, and supporting evidence.
  • The study covers three scenarios involving direct and indirect attacks.
  • The discovered concepts reveal a reusable vulnerability core across models and agents.
  • A frozen VCG requires no further search and outperforms the strongest frozen discovery baseline by 14.2 percentage points under the same single-shot protocol.
  • The resulting VCG provides an auditable artifact for production safety teams to inspect vulnerabilities, validate patches, and accumulate reusable safety knowledge.

Key Concepts Mentioned

  • Automated red-teaming
  • Production language models
  • AHA loop
  • Vulnerability Concept Graph (VCG)
  • Reusable vulnerability core

Claims Made by the Document

  • AHA outperforms the strongest frozen discovery baseline by 14.2 percentage points.
  • The discovered concepts reveal a reusable vulnerability core across models and agents.

Evidence Used

  • Results from experiments on Claude Code and Codex

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Production LLM agents such as Claude Code and Codex operate over untrusted content, files, commands, and workspace state, making safety failures directly actionable. Red-teaming must therefore keep pace with evolving models and tools. Existing approaches mainly optimize attack success and preserve artifacts such as benchmarks, payloads, or attack programs, which record where attacks succeed but not the enabling conditions behind unsafe agent behavior. We study automated red-teaming for production LLM agents using one agentic research environment to discover reusable vulnerability knowledge about another. We present AHA, a falsifiable discovery loop that proposes a vulnerability hypothesis, constructs a falsifier, instantiates a valid attack, executes it in a sandboxed harness, reflects on the trajectory, and promotes confirmed findings into a Vulnerability Concept Graph (VCG). Each concept links an attacker-facing surface to an unsafe trajectory through a claim, enabling condition, falsifier, transfer prediction, and supporting evidence. Across Claude Code and Codex on three scenarios covering direct and indirect attacks, the discovered concepts reveal a reusable vulnerability core across models and agents. A frozen VCG requires no further search and outperforms the strongest frozen discovery baseline by 14.2 percentage points under the same single-shot protocol, while transferring across scenarios and attack channels. The resulting VCG provides an auditable artifact for production safety teams to inspect vulnerabilities, validate patches, and accumulate reusable safety knowledge. Our code is available at https://github.com/henrymao2004/Auto-research-red-teaming-in-sleep.