Evidence Over Plans: Online Trajectory Verification for Skill Distillation

2026-05-09

AI Summary

Existing skill generation methods rely on prior planning and lack empirical evidence from environmental interaction, which compromises skill quality and limits task success rates. This work proposes SPARK, a framework that performs posterior skill distillation from environment interaction trajectories and ensures alignment between generated skills and the task environment through online validation. The core innovations include a trajectory-level measure of skill credibility—termed the Posterior Distillation Index (PDI)—which drives skill diagnosis and targeted intervention. Evaluated across 86 tasks, SPARK-generated skills significantly outperform both no-skill baselines and manually authored skills, while reducing the student model’s inference cost by up to three orders of magnitude.

Abstract

Agent skills can remarkably improve task success rates by using human-written procedural documents, but their quality is difficult to assess without environment-grounded verification. Existing skill generation methods heavily rely on preference logs rather than direct environment interaction, often yielding negligible or even degraded gains. We identify that it is a fundamental timing bottleneck: robust skills should be posterior-based, distilled from empirical environment interaction rather than prior plans. In this study, we introduce the Posterior Distillation Index (PDI), a trajectory-level metric that quantifies how well a distilled skill is grounded in the task-environment evidence. To operationalize PDI, we present SPARK (Structured Pipelines for Autonomous Runnable tasKs and sKill generation) for preserving task execution evidence towards full trajectory-level analysis. SPARK generates environment-verified trajectories used to compute PDI, and it applies PDI as an online diagnostic and intervention signal to ensure posterior skill formation. Across 86 runnable tasks, SPARK-generated skills consistently surpass no-skill baselines and outperform human-written skills on student models (inference cost up to 1,000x cheaper than teacher models). These findings show that PDI-guided distillation produces efficient and transferable skills grounded in the task-environment interaction. We release our code at https://github.com/EtaYang10th/spark-skills .

Problem

• skill distillation
• trajectory verification
• environment-grounded skills
• posterior-based learning
• task execution evidence

Innovation

• Posterior Distillation Index
• Skill Distillation
• Environment-Grounded Verification
• Online Trajectory Verification
• SPARK