Monthly work summary for 2026-03 focusing on automation-driven issue labeling and triage improvements for vllm-ascend. Implemented scalable GitHub Actions workflows and expanded labeling taxonomy to improve issue routing, triage accuracy, and lifecycle management. Key automation delivered includes: (1) automated issue labeling on open/edit with a dedicated issue-labeler workflow and enhanced label taxonomy for model families and core features; (2) expanded and normalized keyword-based labeling rules to improve routing for LLMs, multimodal, audio, and omni scenarios; (3) scheduled stale-issue management to auto-remediate inactive issues by marking stale, reminding, and closing after grace periods; (4) title-based gating for auto-labeling to avoid misclassification; (5) stability and maintainability improvements to CI automation with thorough validation of YAML configurations and regex rules. These changes reduce manual triage effort, improve issue routing consistency, and accelerate issue resolution across the vllm-ascend repository.