Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-03-17 · source-backed
A study of 42,447 agent skills found 14 distinct threat patterns. Existing tools (Semgrep, Bandit) achieved near-zero recall on instruction-level threats. The entire agent skill security gap is invisible to standard DevSecOps. Source
Each link below shares sources, entities, or timing with this story.
SkillScan benchmarked against Bandit / Same source / Shared topic
Linked by a graph relationship (SkillScan benchmarked against Bandit); cite the same source (Source); overlapping topics (agent, pattern, skill).
Shared entity: Existing / Same source domain / Shared topic / What happened next
Both cover Existing; reported by the same outlet (arxiv.org); overlapping topics (agent, skill).
Both cover Existing; reported by the same outlet (arxiv.org); overlapping topics (agent, existing).
Shared entity: Existing / Same source domain / What happened next / Tension
Both cover Existing; reported by the same outlet (arxiv.org); picks up the Existing thread on 2026-07-21.
Both cover Existing; reported by the same outlet (arxiv.org); picks up the Existing thread on 2026-07-17.
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (agent, pattern, skill); pushes against this story (vs).
Shared entity: Existing / Same source domain / What happened next
Both cover Existing; reported by the same outlet (arxiv.org); picks up the Existing thread on 2026-08-17.
Shared entity: Semgrep / Same source domain / What happened next
Both cover Semgrep; reported by the same outlet (arxiv.org); picks up the Semgrep thread on 2026-08-03.