A bidirectional graph alignment method that detects declaration-implementation inconsistencies in LLM agent skills, reaching 87.93% detection F1 on a new 633-skill benchmark.
InProceedingsoftheSixthEuropeanWorkshoponMachine Learning and Systems, 215–224
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SkillConsist: Detecting Inconsistencies in Agent Skills via Bidirectional Graph Alignment
A bidirectional graph alignment method that detects declaration-implementation inconsistencies in LLM agent skills, reaching 87.93% detection F1 on a new 633-skill benchmark.