NESTFUL

Fits agents that must pass results between tools in one plan (data pipelines, calculators, code helpers); borrow its label-and-reference format for nested calls and its metric ladder (API-name F1, parameter F1, partial and full sequence match, executed win rate).

What it asks the agent to do

• write a sequence of API calls where one call's output feeds the next
• name the right function and map parameters to earlier outputs
• solve a math word problem with calculator-style tools
• complete a coding task by composing Python helper functions

What to borrow from it

grading designcase template

Caveats

Only math and coding sources (MathQA, StarCoder2-Instruct), sequences were LLM-generated and filtered by execution, and there is no dialogue or user; the v1 RapidAPI-based subset differs from the v2 set the README now ships.

Details

Links

• Paper
• Github