What problem does it solve?
New feature work in AI coding sessions often loses track of requirements, design decisions, and progress when sessions are interrupted or resumed. This Skill provides a structured workflow that persists task state to project-local files so features can be clarified, implemented, verified, and resumed without losing context.
Core Features & Use Cases
- Full-lifecycle workflow: Covers requirement clarification, design decisions, step-by-step implementation, verification, and a final self-review before archiving.
- Persistent task state: Writes task files to
.codex/tasks/ using a structured template, archives completed tasks to .codex/tasks/archived/, and resumes unfinished tasks instead of restarting.
- Status discipline: Enforces explicit states (需求中, 待开始, 进行中, 已完成) with only one active in-progress task at a time.
- Use Case: You ask the AI to build a new filtering feature, get interrupted mid-implementation, and later say "continue the last task" — the Skill scans
.codex/tasks/, finds the unfinished task file, and resumes from the recorded state.
Quick Start
Ask the AI to use cc-new-feature to plan and implement a new feature, persisting progress in the project's .codex/tasks directory.
