SYNTHESIS NOTE
Topics›Agents Multi Architecture›this note

Can attackers evade skill scanners by refining individual skills?

Explores whether feedback from per-skill scanners can be weaponized to make malicious multi-skill chains undetectable. Matters because it tests a core assumption of skill-level defense mechanisms.

Synthesis note · 2026-09-23 · sourced from Agents Multi Architecture

The paper pairs two components. "LLM-based chain planning" decides how the intent is decomposed into sub-skills and in what order they hand off. "Scanner-feedback refinement" then works on the individual sub-skills, "iteratively reducing suspicious signals" while preserving "chain-level attack semantics." The reported result is an average ASR of 96.0 percent across six representative skill scanners, the best among the evaluated baselines.

Why the two components fit together is my reading, not the paper's. A skill scanner scores one skill at a time, so the only thing its feedback can teach is how to make one skill look blander. The attacker can follow that direction without touching the chain, because the chain's meaning sits in the planner's decomposition, a level no per-skill score reaches. The defender's output then works as a search signal aligned with the attack: each round makes the pieces less suspicious and leaves the composed behavior intact. What makes detecting AI agent traps fundamentally difficult? expects attackers to probe and work around each defense. This is a sharper form of that expectation, because the scanner's own report is the probe. A formal cousin sits in an idealised setting: Can repeated quiet probes separate decoys from genuine objects? says enough quiet probes separate decoys from genuine objects once their response distributions differ and can be learned. The likeness is loose. That result is a theorem over a fixed-candidate benchmark and this one is an empirical attack on scanners, and neither excerpt connects them.

The strongest objection is about what 96.0 measures. Refinement against a scanner's feedback is the condition under which that scanner should fail most, so the figure may say more about an attack tuned to these six scanners than about how an untuned chain fares. The excerpt does not report a pre-refinement rate.

The paper adds that the chain "executes successfully on OpenCode, Claude Code, and Codex with different model backbones." I read this as evidence that the attack works through the skill mechanism instead of the quirks of one agent, but it is the less specified of the two results: no count or rate per runtime is given.

What the excerpt does not give. Scanner names, per-scanner rates, the baselines, the number of chains, and a definition of ASR (evading the scanner, executing the payload, or both). Without that definition the 96.0 and the execution claim are not on a common footing.

Inquiring lines that read this note 115

This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.

How prevalent is reward hacking in frontier models? How can defenders detect coordinated attacks across episodes? How does outcome-only reporting obscure which system components blocked attacks? Can defenses detect attacks composed across multiple skills? Where do unmonitored channels leave multi-agent planning vulnerable to attack? Do current AI defenses adequately protect against semantic manipulation attacks? How can honeytokens stay effective against compromised insider threats? Do multi-agent systems create greater security risks than single-agent ones? How do training data composition and selection affect model capabilities? How can evaluations detect conditional compliance in monitored AI systems? What determines whether AI output can be epistemically verified and trusted? How can workflow-level validation detect semantic corruption that protocol compliance misses? Does situational awareness enable models to exploit evaluation gaps? What conditions enable agent collusion in multi-agent verification tasks? How does position in multi-agent workflows amplify or attenuate harmful signals? What infrastructure evidence validates agent benchmark achievement claims? How do models reward hack during evaluation and can detection succeed? How do persistent skill repositories improve agent reliability over time? How does training for improved reasoning reduce abstention ability? What determines whether AI system errors remain visible and contestable? How do evaluation methodologies affect which model capabilities are revealed or hidden? What causes model scheming and how do we distinguish it from accidents? How reliable are reasoning traces as evidence of agent honesty? How can we verify agent claims against their actual capabilities and actions? Can strategic routing of diverse smaller models outperform a single scaled model? Do frontier models develop hidden self-protective behaviors? How do agents balance task completion with privacy compliance and security? How do coordinated agent sequences violate constraints that individual actions respect? Do planted honeypot tests reliably measure reward hacking? How does training data contamination persist through safety alignment mechanisms?

Related concepts in this collection 5

This note in its neighbourhood — explore the map, then jump to a related concept in the list below.

Concept map
15 direct connections · 105 in 2-hop network ·medium cluster Open in graph ↗

Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph

your link semantically near linked from elsewhere

Related papers in this collection 8

Papers most semantically related to this note, ranked by cosine similarity in the embedding space.

Original note title

scanner feedback lets an attacker blunt each sub-skill while chain planning holds the attack together — ColluSkill reaches 96 percent average attack success across six skill scanners