auto-dev-orchestrator

Category: Docs & Knowledge | Uploader: liutiger | Downloads: 0 | Version: v1.0（Latest）

Use only when implementation has been approved and the next step is a full end-to-end delivery loop. Not for early planning, standalone review, incident triage, or documentation-only work.

Changelog: Source: GitHub https://github.com/liutiger/ai-collab-skill-playbook

Directory Structure

Current level: template/.claude/skills/wms/auto-dev-orchestrator/

📁 agents/
- 📄 openai.yaml 314 B
📁 references/
- 📄 checklists.md 385 B
- 📄 output-contract.md 503 B
- 📄 workflow.md 699 B
📄 SKILL.md 1.1 KB

SKILL.md

---
name: auto-dev-orchestrator
description: "Use only when implementation has been approved and the next step is a full end-to-end delivery loop. Not for early planning, standalone review, incident triage, or documentation-only work."
---

# Auto Dev Orchestrator

## When to Use

- 复杂任务需要 AI 走完整闭环
- 跨模块需求不能只停在分析阶段
- 需要把实现、自检和落盘收紧在同一次协作中

## Pair With

- `wms-task-governance`
- `gitnexus-code-navigation`
- `delivery-evaluation-gate`
- 对应场景薄 Skill：`feature-dev-scene`、`issue-investigation-scene`、`code-review-scene`、`database-change-scene`、`refactoring-scene`、`documentation-scene`

## Workflow

1. 读 [workflow](references/workflow.md)
2. 按 [output contract](references/output-contract.md) 输出阶段结果
3. 实施完成后，逐项核对 [checklists](references/checklists.md)

## Boundaries

- 不要跳过 Planner / Context / Architect 阶段直冲实现
- 不要忽略场景薄 Skill 的专项检查
- 不要把“已改代码”当成“已完成闭环”
- 不要在未经过评测门禁时默认任务已完成

Comments 0

Latest | Hot

Please login before commenting.

No comments yet. Be the first one!

auto-dev-orchestrator

Directory Structure

SKILL.md

Report

Notice