NexForge: Scaling Executable Agent Tasks via Requirement-First Synthesis

2026-07-15

AI Summary

This work addresses the limitations of existing tool- or skill-graph-based approaches for generating agent training data, which are constrained by predefined bases and fail to capture the full distribution of real-world tasks. To overcome this, the authors propose a requirement-driven framework for synthesizing executable agent training data, leveraging requirement mining, distribution-aware task compilation, and automated environment construction to enable cross-domain, unified data generation without reliance on pre-specified primitives. The method produces 3,600 terminal tasks and 2,000 office tasks, boosting model performance on Terminal-Bench 2.0 from 22.5% to 52.0% and enabling the Nex-N2 series models to achieve state-of-the-art results among open-source alternatives.

Abstract

Scaling executable agent training data is bottlenecked by substrate-first methods that tie task generation to predefined tools, repositories, or skill graphs: expanding coverage requires manual expansion of the substrate, each new domain demands a bespoke pipeline, and the resulting task distributions often reflect substrate convenience rather than real-world demand. We introduce NexForge, a requirement-first framework that compiles free-form capability requirements into executable agent training data. NexForge first performs research-based demand discovery to identify representative task forms, realistic scenarios, and their relative prevalence. It then applies distribution-aware task compilation and automatically retrieves or constructs the files, repositories, dependencies, and runtime configurations required to materialize each task, followed by teacher rollout collection and trajectory distillation. The same pipeline, without any domain-specific infrastructure, produces 3,600 terminal tasks and 2,000 office tasks, improving Qwen3.5-35B-A3B Base from 22.5% to 52.0% on Terminal-Bench 2.0 and from 813 to 1338 Elo on GDPval; scaling to 43.2K terminal tasks reaches 58.4%, surpassing Claude Opus 4.6. Scaled further, NexForge-synthesized data contributes to the training of Nex-N2, a family of publicly available agent models that lift Qwen3.5-35B-A3B to 75.3% on Terminal-Bench 2.1 and to 1585 Elo on GDPval -- achieving state-of-the-art open-source performance and surpassing several frontier proprietary systems. Nex-N2 models are available at https://nex.sii.edu.cn/

Problem

• executable agents
• task generation
• training data scaling
• requirement-first synthesis
• real-world demand

Innovation

• requirement-first synthesis
• executable agent
• task generation
• distribution-aware compilation
• trajectory distillation