SkillFuzz frames implicit-intent discovery in open agent skill marketplaces as a fuzzing problem over skill compositions rather than isolated skills. It extracts structured skill contracts, uses contract-guided Monte Carlo Tree Search to prioritize potentially conflicting combinations, and inspects planning artifacts before execution. A skill-free baseline provides a differential oracle for detecting deviations in agent intent. Across representative marketplace workloads, the paper reports more than 1,000 distinct implicit intents under a fixed query budget, confirmation of over 80% of the highest-risk flagged compositions during execution-time validation, and better discovery of high-severity intents than alternative search strategies while exploring only a fraction of their pairwise interaction space.
No heat snapshots are available in the last 24 hours.