"Instruction Bleed" names a structural vulnerability in prompt-composed agents.

Same source domain / Shared topic / Tension

Auto-mined agent skills are interpretable but barely move the needle.

Reported by the same outlet (arxiv.org); overlapping topics (agent, another, skill); pushes against this story (against).

"Strained coherence" might be the early-warning metric for agents going off the rails.

Reported by the same outlet (arxiv.org); overlapping topics (agent, author, behavior); pushes against this story (against).

Architecture Descriptors Cut AI Coding Agent Navigation by 33-44%: 7,012 Claude Code Sessions Analyzed

Reported by the same outlet (arxiv.org); overlapping topics (agent, boundary, module); pushes against this story (but).

AgentSentry — Temporal Causal Defense Against Prompt Injection

Reported by the same outlet (arxiv.org); overlapping topics (agent, behavior, boundary); pushes against this story (against).

Same source domain / Shared topic

A paper on procedural memory asks whether agents actually reuse skills or just accumulate cruft.

Reported by the same outlet (arxiv.org); overlapping topics (actually, agent, behavior, skill).

Knowledge Activation: AI Skills as the Institutional Knowledge Primitive

Reported by the same outlet (arxiv.org); overlapping topics (agent, behavior, composable, skill).

Shared entity: Treat / Shared topic / Earlier coverage

Three AI coding agents leaked secrets through one prompt injection payload.

Both cover Treat; overlapping topics (agent, boundary); earlier Treat coverage from 2026-06-19.

Shared entity: Treat / Same source domain / Earlier coverage

Fine-tuned vuln detectors may be pattern-matching, not reasoning.

Both cover Treat; reported by the same outlet (arxiv.org); earlier Treat coverage from 2026-06-19.