[research@ai4se] : ~ $ ▌

cat ~/ai4se/README.md

# AI 时代的软件工程分层笔记

## Methodology • Process • Tools • Harmony • Measurement

围绕 AI4SE / Agentic Engineering，记录软件工程在方法、流程、工具、人机协作与效能度量上的系统性变化。

cat ~/notes/ai4se-layers.md | ls ~/notes/*/
research@ai4se:~ — outline
$ tree ~/notes -L 2
[OK] methodology/ spec-driven/ agentic-engineering/
[OK] process/ workflow-design/ review-pipeline/
[OK] tools/ mcp-protocol/ harness-engineering/
[OK] harmony/ hitl-hotl/ governance/
[OK] measurement/ dora-metrics/ dev-productivity/
 
$ grep -r "Effectiveness" ~/notes
→ 15 draft articles indexed
$ npm run build
[OK] static pages generated from Markdown
 
$ ▌
markdown taxonomy ci-ready

man -k ai4se

(1) METHODOLOGY —

方法论 — SDD、Agentic Engineering、AI4SE 体系

ls ~/notes/methodology/
(2) PROCESS —

流程 — 人机协作工作流、Review Pipeline、Loop 设计

ls ~/notes/process/
(3) TOOLS —

工具 — MCP 协议、Harness Engineering、浏览器自动化

ls ~/notes/tools/
(4) HARMONY —

协作 — HITL/HOTL/HOOL、Planner/Generator/Evaluator

ls ~/notes/harmony/
(5) MEASUREMENT —

度量 — DORA、SPACE、DevEx、Agent 效能

ls ~/notes/measurement/

grep -r "featured" ~/notes/

02

AI4SE 端到端软件交付最佳实践

2025–2026 行业共识：AI 是放大器而非万能药——端到端成功依赖规格→上下文→执行→验证→度量的闭环，而非 IDE 内补全。

[methodology] [e2e-delivery][methodology] | 2026-06-21 | 16 min

dpkg -l | grep ^i

[research@ai4se ~]$ cat /etc/software-engineering.layers
Methodology
layer 1 SDD / Agentic Engineering
Process
layer 2 Loop / Pipeline / Review
Tools
layer 3 MCP / Harness / Automation
Harmony
cross-cut HITL / HOTL / HOOL
Measurement
feedback DORA / SPACE / DevEx
Effectiveness
foundation Quality + Efficiency

# 内容目录即知识库：只要新增 / 修改 Markdown，GitHub Actions 就会重新生成静态博客。