Evidence · DiggingBeagle record
SkillLifeBench: skill lifecycle decisions and stale installed execution
Benchmark of agent skill updates, version selection and lingering loaded artifacts with task and attacker outcome distinctions.
- Published
- Oct 29, 2026
- Source role
- primary disclosure
Evidence record
Benchmark of agent skill updates, version selection and lingering loaded artifacts with task and attacker outcome distinctions.
Claim-level citations (1)
- supportsAn agent selected a patched skill but executed an old loaded handle in a benchmark: SkillLifeBench documents a controlled case in which an agent chose patched skill v1 but executed a stale v0 handle with an unsafe filesystem capability.
SkillLifeBench paper, version-update and stale-handle experiment