🚀 STDD V2.9.3 正式发布 STDD V2.9.3 Released

[leonai42]

---

中文版

🚀 STDD V2.9.3 正式发布：三档执行模式 + beyond code 扩展 + 白皮书一套四本

STDD（Spec+Test Driven Development）V2.9.3 正式发布！这是自 V2.5 以来最大的一次版本升级，核心思路很简单：让同一套方法论，同时覆盖"改个
typo"和"重构架构"两种极端场景。

🆕 V2.9 六大新特性

⚡ 三档执行模式

Phase 1 末尾自动计算 6 维度复杂度评分（0-17 分），智能推荐执行模式：

模式	评分	适用	特点
🪶 轻量	0-3	bug 修复、小调整	简化 SPEC → 跳过 SLICE → RED → GREEN
⚖️ 标准	4-7	功能增强、中等重构	完整六阶段流程
🔬 彻底	8-17	新模块、架构变更	标准 + pass@k + 安全/性能子代理

核心原则不变：任何模式下，TDD 底线（RED→GREEN）不可跳过。

🎯 智能门禁系统

四级范围分类器（micro / small / medium / large），自动分析变更范围给出建议，告别一刀切的"允许/阻止"。

📦 批次目录管理

轻量模式下微小修复共用批次目录，按时间窗口自动归档。新增 stdd batch CLI 命令族。

🔄 版本升级管理

stdd upgrade 一键全局升级，upgrade --check 差异检查，upgrade --force 强制同步。

🔬 Plankton + pass@k

三级自动修复（格式化/建议/报告）+ 同一测试运行 k 次的统计验证，用于检测 Spec 歧义。

🌐 STDD beyond code

V2.9 最具突破性的升级：从编程方法论扩展为通用质量保障方法论。只要可定义"预期结果"和"验证方式"，任何任务 — 量化策略审核、合规校验、内容发布检查 — 都适用。

---

📊 V2.5 → V2.9 量化对比

指标	V2.5	V2.9	变化
CLI 命令	18	28	+56%
测试用例	155	250	+61%
源码行数	~5,400	~10,800	+100%
执行模式	1 种	3 种	轻量/标准/彻底
失败模式检查	11 类	12 类	+锚定缺失
AI 平台	6	7	+OpenCode

---

📚 白皮书一套四本 同期发布

版本	说明
📘 中文版	面向中文读者的方法论完整阐述
📙 English	For the international community
🤖 中文 · For AI	结构化、可机读，专为 AI Agent 消费设计
🧠 EN · For AI	Structured, machine-readable for AI agent consumption

📥 下载方式：官网 stdd.ai 下载专区，或 GitHub Releases。

---

🔗 相关链接

• GitHub 仓库：https://github.com/leonai42/stdd
• 官网：https://stdd.ai
• 更新日志：https://github.com/leonai42/stdd/blob/master/CHANGELOG.md
• 框架深度对比：https://github.com/leonai42/stdd/blob/master/AI_CODING_METHODOLOGY_COMPARISON.md

---

质量不靠运气，靠系统。Spec 先行，TDD 执行。

欢迎试用、反馈、贡献 🙌

---

English Version

🚀 STDD V2.9.3 Released: 3 Execution Modes + Beyond Code + 4-Book Whitepaper Set

STDD (Spec+Test Driven Development) V2.9.3 is officially released! This is the largest upgrade since V2.5, built around a simple idea: one methodology that covers
both "fix a typo" and "refactor the architecture."

---

🆕 Six Major New Features in V2.9

⚡ Three Execution Modes

A 6-dimension complexity score (0-17) is auto-calculated at the end of Phase 1:

Mode	Score	Use Case	Flow
🪶 Light	0-3	Bug fixes, small tweaks	Simplified SPEC → skip SLICE → RED → GREEN
⚖️ Standard	4-7	Features, moderate refactors	Full 6-phase flow
🔬 Thorough	8-17	New modules, architecture changes	Standard + pass@k + security/perf sub-agents

Invariant: TDD baseline (RED→GREEN) is never skipped, in any mode.

🎯 Smart Gate System

A 4-level scope classifier (micro / small / medium / large) provides intelligent recommendations instead of binary allow/block.

📦 Batch Directory Management

Lightweight changes share batch directories, auto-archived by time window. New stdd batch CLI command family.

🔄 Version Upgrade Management

stdd upgrade for one-click global upgrades, --check for diff inspection, --force for sync.

🔬 Plankton + pass@k

3-level auto-fix (format/suggest/report) + statistical verification (k runs detect Spec ambiguity).

🌐 STDD Beyond Code

V2.9's most breakthrough upgrade: from a coding methodology to a universal quality assurance methodology. Any task with definable expected results and verification methods — quant strategy review, compliance checks, content publishing — can benefit.

---

📊 V2.5 → V2.9 By the Numbers

Metric	V2.5	V2.9	Delta
CLI Commands	18	28	+56%
Test Cases	155	250	+61%
Source Lines	~5,400	~10,800	+100%
Execution Modes	1	3	Light/Standard/Thorough
Failure Checks	11	12	+Anchor Missing
AI Platforms	6	7	+OpenCode

---

📚 4-Book Whitepaper Set — Released Together

Edition	Description
📘 中文版	Full methodology for Chinese readers
📙 English	Full methodology for the international community
🤖 中文 · For AI	Structured, machine-readable for AI agent consumption
🧠 EN · For AI	Structured, machine-readable for international AI agents

📥 Download: stdd.ai or GitHub Releases.

---

🔗 Links

• GitHub Repo: https://github.com/leonai42/stdd
• Website: https://stdd.ai
• Changelog: https://github.com/leonai42/stdd/blob/master/CHANGELOG.md
• Framework Comparison: https://github.com/leonai42/stdd/blob/master/AI_CODING_METHODOLOGY_COMPARISON.md

---

Quality by system, not by luck. Spec first, TDD execution.

Try it out, share feedback, and contribute 🙌

---