Moyu Front-end Platform: Architecture, Challenges, and Solutions

After nearly eight years of growth, Xianyu's front‑end has become a critical part of its R&D system, offering fast iteration, strong dynamism, and low cross‑platform adaptation cost. To meet increasing business complexity, the team launched the codename “Moyu” in late 2022 as a dedicated front‑end R&D framework.

Platform Overview – Moyu consists of three major parts: a production‑research platform, a development model, and a gateway construction.

Production‑Research Platform – Previously, Xianyu relied on the group‑wide engineering platform for app creation, Git repository management, and build/deploy pipelines. The workflow involved creating an app, creating an iteration, pulling code locally, committing, building H5 assets ( CSS/JS/HTML), deploying to a web server, testing, and finally releasing. This process depended heavily on the group platform, which lacked flexibility for complex, customized needs.

Problem Analysis – Issues included scaffold and configuration difficulty, high manual effort for new developers, and fragmented page‑and‑data release processes.

Solutions – Moyu introduces unified template management, integrated configuration management, scenario‑based page management, and a closed‑loop release process, eliminating the need for separate code maintenance of frameworks.

Development Model – The legacy H5 development offered two modes: source‑code development and module assembly. As business grew, a hybrid approach was needed. Moyu now supports a template‑driven mode where the final output is XTPL/JS/CSS/Schema. XTPL is a dynamic document template cacheable by CDN, and Schema describes configuration data for the editor.

This model enables “one‑code‑multiple‑pages” and independent data‑driven deployments, improving flexibility and reducing redundant UI components.

Gateway Construction – To address first‑screen performance and data‑dependency issues, Moyu implements a lightweight front‑end data gateway that consolidates multiple data calls, reduces serial requests, and standardizes common data models (e.g., product and content). The gateway improves first‑contentful‑paint (FCP) and cross‑platform first‑screen times, as shown by the performance comparison on the “Mobile Digital” channel.

Current Status & Future Plans – Moyu is being applied to standard data‑model scenarios such as product recommendation and content feeds. Future work includes extending support to mini‑programs and high‑performance containers, enriching data‑targeting capabilities (AB testing, audience segmentation), integrating with other internal platforms (e.g., monitoring, testing), and offering the Schema‑based configuration editor as a product for other Xianyu back‑office systems.