Get Started
Research questionHow can cold-start LLM agents reliably improve an initial procedural skill?An initial skill may be syntactically valid yet fail during execution, while expert authoring is costly and one-shot generation may not reflect how agents actually perform tasks. The central difficulty is improving the skill when accumulated self-evolution trajectories are unavailable.
AI
AI Agents
Evaluation & Benchmarks
Latest papersRecent research connected to this question, newest first.SkillRevise: Improving LLM-Authored Agent Skills via Trace-Conditioned Skill RevisionThe source concerns procedural skills that help LLM agents execute workflows, verify constraints, and recover from failures. Its evidence covers three benchmarks, two domain-specific studies, and six LLMs, including transfer across executors and task environments; it does not establish broader deployment capabilities beyond those settings.research paper · Sep 3, 2026
Related questions
How can LLM agents reuse procedural skills across diverse long-horizon tasks without generic or instance-bound memory?How can LLM agents jointly adapt reasoning policies and hierarchical skill libraries during reinforcement learning?How can agent runtimes avoid context poisoning and latency from growing histories during long-horizon skill execution?How can LLM agents retrieve relevant skills from large, noisy libraries under context and latency constraints?
Home
Topics
Search
Library