Get Started
Home
Topics
Search
Library
Research questionHow can open-source LLM agents reliably chain multiple tool calls over live public APIs?Open-source LLM agents often fail when one API call must provide the inputs for later calls. These failures make real-world public-sector tasks difficult to execute reliably.
AI
AI Agents
Evaluation & Benchmarks
LLM Pretraining & Post-training
Reasoning
Technology
Latest papersRecent research connected to this question, newest first.Multi-Step Tool-Calling over Korean Open Public APIs: A Benchmark and a Data-Synthesis RecipeThe source focuses on on-premise agents using Korean government APIs and reports evidence from 145 KOPA-Bench tasks, with additional results on BFCL. It does not establish performance across other API ecosystems or deployment settings.research paper · Sep 4, 2026
Related questions
How can we train and evaluate LLM agents for tool use across single- and multi-turn workflows with serial or parallel calls?How can LLM agents reconcile conflicting user instructions, stored knowledge, and live tool observations before acting?How can tool-using agents prevent sensitive conclusions assembled from individually non-revealing tool outputs?How can LLM-agent systems prevent safety compromises from propagating across workflow boundaries?