• Computer Science > Artificial Intelligence [Submitted on 23 Feb 2026] Title:Learning to Rewrite Tool Descriptions for Reliable LLM-Agent Tool Use View PDF HTML (experimental)Abstract:The performance of LLM-based agents depends not only on the agent itself but also on the quality of the tool interfaces it consumes. • While prior work has focused heavily on agent fine-tuning, tool interfaces-including natural language descriptions and parameter schemas-remain largely human-oriented and often become a bottleneck, especially when agents must select from large candidate tool sets. • Existing approaches to improving tool interfaces rely on execution traces, which are frequently unavailable in cold-start or privacy-constrained settings, and typically optimize each tool independently, limiting scalability and generalization to unseen tools. • We propose Trace-Free+, a curriculum learning framework that progressively transfers supervision from trace-rich settings to trace-free deployment, encouraging the model to abstract reusable interface-usage patterns and tool usage outcomes. • To support this approach, we construct a large-scale dataset of high-quality tool interfaces using a structured workflow over a diverse collection of tools. • Experiments on StableToolBench and RestBench show consistent gains on unseen tools, strong cross-domain generalization, and robustness as the number of candidate tools scales to over 100, demonstrating that tool interface optimization is a practical and deploya

Article Summaries:

  • Researchers have introduced Trace‑Free+, a curriculum‑learning framework that improves the natural‑language descriptions and parameter schemas of tools used by large‑language‑model (LLM) agents. By progressively transferring supervision from trace‑rich to trace‑free settings, the method encourages models to learn reusable interface‑usage patterns without relying on execution logs, which are often unavailable. The team built a large dataset of high‑quality tool interfaces through a structured workflow over diverse tools. Experiments on StableToolBench and RestBench show consistent gains on unseen tools, strong cross‑domain generalization, and robustness even when the candidate tool set exceeds 100, demonstrating that optimizing tool interfaces is a practical complement to agent fine‑tuning.

Sources: