Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add TashanGKD/tashan-cursor-skills --skill engineering-principle-recordergit clone --depth 1 https://github.com/TashanGKD/tashan-cursor-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/engineering-principle-recorder)<a href="https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/engineering-principle-recorder"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-cursor-skills/engineering-principle-recorder/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/engineering-principle-recorder"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-cursor-skills/engineering-principle-recorder.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00118 | $0.03919 |
| Opus 5 | $0.00059 | $0.01959 |
| Sonnet 5 | $0.00024 | $0.00784 |
| Haiku 4.5 | $0.00012 | $0.00392 |
Grade A, and why
engineering-principle-recorder scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 267 lines — stays where its author put it; the contents beside it link to each section on GitHub.
工程/执行原则录入器(engineering-principle-recorder)
核心目标:一条原则被说出来 → 两周后,另一个 AI session 自动被它约束。 这不是「记在本子里」,而是「写入 AI 每次必读的执行层」。
认知根
认知根:_内部总控/认知结构/L1_系统性文档/系统架构思维维度/Skill体系设计原则_v1.0.md
关联:本 Skill 维护「Skill 体系执行约束」的完整性,与设计原则中「执行层」概念直接对应。
两类原则的本质区别(AI 执行者必须先理解)
| 类型 | 内容 | 走哪里 |
|---|---|---|
| 认知原则 | 郑总如何思考/判断(「先从小点出发升维」) | → cognitive-extract-principle → L1.5 |
| 执行原则 | AI系统/代码应如何运行(「LLM调用不截断」) | → 本 Skill → Role Skill / Rule / 追踪台 |
知识导航表(执行前按顺序读取)
| 层级 | 文档 | 读取目的 |
|---|---|---|
| D0 | _内部总控/元项目导航.md |
提取当前活跃项目路径,用于 Step 2-D 的代码搜索目录和追踪台路径 |
| ① | .cursor/skills/skill-index/SKILL-INDEX.md |
确认 Step 2-C 中相关 Role Skill 的实际文件路径 |
激活后立即执行
Step 0 【读取必要上下文】(静默完成,不向用户输出)
Read: _内部总控/元项目导航.md
→ 记录当前活跃项目名称 + 代码路径 + 追踪台路径
Read: .cursor/skills/skill-index/SKILL-INDEX.md
→ 记录 Role Skill 的实际文件路径
Step 1 【精确化原则 + 类型分类(必须与用户交互完成)】
⚠️ 执行铁律:先确认原则精确表述 + 分类,再执行写入。禁止跳过。
⚠️ 分类完成后,无需用户再次指示,立即执行对应的 Step 2。
Step 1-1 精确化原则表述
向用户确认:「我理解您要立的原则是:[提炼后的精确表述]。
判断标准是:[什么情况算违反了此原则]?」
→ 等待用户确认后,以确认版本作为写入内容,不用原话
Step 1-1.5 【等效三问声明(A1.9 门禁前置满足)】
本 Skill 在 Step 1 已等效完成 skill-rule-修改规范 三问:
① 根因 = Step 1-1 用户确认的原则内容(为什么要改)
② 影响范围 = Step 1-2 分类后确定的写入目标文件
③ 验证方法 = Step 3 覆盖报告中的可靠性评估
→ AI 执行时在 Step 1 末尾内部记录:「A1.9 三问已完成,可继续写入 .cursor/ 文件」
→ 不需要再次触发 skill-rule-修改规范 的完整流程
Step 1-2 类型分类
─── 类型 A:认知/思维原则 ───
判据:原则描述的是「郑总如何思考/判断/分析问题的方式」
示例:「我倾向于先从小点出发」「推断前先验证」
处理:告知用户「这条原则描述的是您的思维方式,应通过
cognitive-extract-principle 流程录入到 L1.5 底层原则库。
本 Skill 在此停止——请您说『提炼原则』以激活该流程。」
→ 本 Skill 在此终止,不执行 Step 2 和 Step 3。
─── 类型 B:AI 行为约束 ───
判据:原则描述的是「AI 执行任务时如何使用工具/读文档/调用资源」
(不涉及代码设计,只约束 AI agent 的操作方式)
示例:「读文档不截断,分次读但不遗漏」
「连接服务器前先确认是哪台机器」
→ 标记 B = true,进入 Step 2-B
─── 类型 C:工程设计原则 ───
判据:原则描述的是「代码/LLM调用/系统架构应如何设计」
示例:「LLM 调用上下文不截断,qwen3.5plus 256K 够用」
「认证必须复用现有服务,不得重新实现」
→ 标记 C = true,进入 Step 2-C
─── 类型 D:当前代码存在违规 ───
判据:用户明确指出「现在代码里已有违反这条原则的地方」
OR 本 Skill 执行 Step 2-C 后判断极可能有历史违规
→ 标记 D = true,在 Step 2-C 完成后进入 Step 2-D
⚠️ 多类型并发:B+D 和 C+D 都是合法组合,逐类处理,不跳过任何一类
⚠️ 若分类不确定 → 问用户:「这条原则是关于AI的操作方式(B),
还是代码的设计方式(C),还是两者都是?」
─── 与相似 Skill 的区分 ───
「这里有个规范问题/这次违反了规范」(已有规范被违反一次)→ project-convention-resolver
「要立一条新原则」(原则本身尚不存在)→ 本 Skill
若不确定 → 问用户:「是要修复这次违反,还是建立一条新规范?」
Step 2-B 【AI 行为约束 → 写入执行层(B类)】(仅当 B = true)
⚠️ 必须实际工具调用写文件,禁止只口头描述「已录入」
B-1 判断写入目标(判据如下)
普遍约束(对所有角色/所有会话均生效)→ session-bootstrap.mdc
特定角色约束(仅当某 Role Skill 激活时生效)→ 对应 Role Skill 的约束节
判断标准:「去掉所有角色限定词,这条原则还成立吗?」
→ 是 → 普遍约束 → session-bootstrap
→ 否 → 特定角色约束 → 对应 Role Skill
B-2 Read 目标文件(必须先读后写)
B-3 按 skill-rule-修改规范 备份
Shell: cp [目标文件] [history/目录/文件名_vX_YYYYMMDD_before_principle.md]
B-4 StrReplace 写入
→ 在目标文件「约束」节或等效节末尾追加:
`- [原则标题]:[精确描述](来源:engineering-principle-recorder,[日期])`
B-5 StrReplace 追加变更记录到文件末尾
B-6 输出:「[B类] 已写入 [文件路径] / [节名称]」
→ 完成后无需等待,继续检查是否有 D = true
Step 2-C 【工程设计原则 → 写入 Role Skill 第一性原理(C类)】(仅当 C = true)
⚠️ 必须实际工具调用写文件,禁止只口头描述「已录入」
C-1 识别最相关的 Role Skill(可多选,用 SKILL-INDEX.md 确认路径)
LLM/AI调用/Prompt → `.cursor/skills/role-AI工程师/SKILL.md`
部署/运维/服务器 → `.cursor/skills/role-DevOps/SKILL.md` 或 `REFERENCE.md`
前端实现 → `.cursor/skills/role-前端开发/SKILL.md`
后端实现 → `.cursor/skills/role-后端开发/SKILL.md`
技术架构 → `.cursor/skills/role-技术架构师/SKILL.md`
所有角色通用 → 同时写 session-bootstrap(B类处理)
C-2 Read 目标 Skill 文件
→ 找「第一性原理」节或等效约束节
→ 若无该节 → 在 Skill 文件正文开头(---分隔符之后)新建「## 第一性原理」节
→ 检查是否有语义相近的已有原则(如有矛盾 → 暂停,告知用户)
C-3 按 skill-rule-修改规范 备份 + StrReplace 写入 + 追加变更记录
写入格式:`- [原则标题]:[精确描述](来源:engineering-principle-recorder,[日期])`
C-4 输出:「[C类] 已写入 [文件路径] / [节名称]」
→ 完成后无需等待,继续检查是否有 D = true
→ ⚠️ 若任一目标文件写入失败:
- 在 Step 3 报告中标注「未覆盖:[文件路径](写入失败)」
- StrReplace 追加到 PENDING-EXPERIENCES.md:
`| [今日] | engineering-principle-recorder | 步骤偏差 | 原则[名称]写入[文件]失败,需补录 | 🔲 待处理 |`
- 不静默跳过,不假装写入成功
Step 2-D 【代码违规审计 + 追踪(D类)】(仅当 D = true)
D-1 构建搜索模式(必须至少 3 个候选词,覆盖不同代码写法)
基于原则语义推导代码中可能的违规模式,例如:
「不截断」原则 → 搜索词:`[:6000]` `[:3000]` `[:500]` `max_tokens=` `content[:`
「必须复用认证」原则 → 搜索词:`jwt.encode` `create_token` `def authenticate`
⚠️ 不同原则需不同搜索词,必须基于原则语义推导,不可照抄示例
D-2 执行代码搜索
Grep: [当前项目代码目录](从 Step 0 的元项目导航获取路径)
→ 找出语义违规位置(非机械匹配,需要 AI 判断)
→ 区分「日志打印截断」([:120] 等,无害)vs「LLM输入截断」(有害)
D-3 写入技术问题追踪台
Read: [项目路径]/技术架构师/技术问题追踪台.md
→ 对每个违规位置,StrReplace 追加一条 issue:
格式:| P1 | [日期] | [文件路径:行号] | 违反「[原则名]」:[具体描述] | 🔲 待修复 |
D-4 输出违规清单给用户:
「发现 N 处违规(已记录到追踪台):
1. [文件:行号] — [描述]
...
是否立即触发 bug-fix-loop-coordinator 修复?」
⚠️ 等待用户回复后继续执行 Step 3(无论用户如何回答,Step 3 必须执行)
Step 3 【输出覆盖报告】(所有工具调用完成后才执行,禁止提前输出)
⚠️ Step 3 是强制步骤,无论走了哪些 Step 2,都必须执行
输出格式:
────────────────────────────────────────────
✅ 原则录入完成
原则名称:[提炼后的标题]
原则内容:[精确描述,含判断标准]
执行层覆盖:
[B类] [文件路径] / [节名称]:alwaysApply → 每次会话强制生效(高可靠)
[C类] [文件路径] / [节名称]:角色激活时必读(中高可靠)
[D类] 发现 N 处历史违规 → 已记录到 [追踪台路径]
可靠性评估:
- [当前录入方式] → [可靠性等级]
- 如需最高可靠性建议:[具体建议,如「也写入 session-bootstrap」]
────────────────────────────────────────────
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 267 lines · 118 tokens per session scan A 9b39c8622381
engineering-principle-recorder is a skill published in the GitHub repository TashanGKD/tashan-cursor-skills (20 stars, last pushed 5mo ago), licensed MIT. It adds 118 tokens to every session and 3,919 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
android-feature
Workflow for implementing a new Android feature, screen, fragment, activity, dialog, adapter, or enhancement in this XML/MVVM template. Use whenever the user asks to add, build, implement, or extend functionality — before writing any code. Covers planning, the MVVM/MVI skeleton, the Android configuration checklist…
template-helpers
Catalog of the reusable building blocks shipped with this Android template — BaseActivity/BaseFragment, the helpers/ extension files (navigation, lifecycle, toast, snackbar, dialogs, images, permissions, theme, locale, settings intents, date, delay), common/ (Firebase, network, observers), and…
android-new-project
Workflow for turning a fresh copy of this Android template into a new app — renaming the package/namespace/applicationId, branding (icons, splash, palette, fonts), Firebase setup, signing config, stripping unused sample screens, and verifying the foundation. Use when the user says they are starting a new app…
android-preflight
Final verification checklist to run before declaring Android work finished — build, both themes, string resources, lifecycle and leak risks, registered permissions and components, resource parity between values and values-night, and honest reporting of what was and was not verified. Use at the end of any feature, fix…
android-ui-theming
Rules for Android XML layouts, Material 3 components, semantic color tokens, light/dark theming, dimensions (sdp/ssp), string resources and localization, accessibility, and edge-to-edge insets. Use when creating or editing any layout, drawable, style, color, vector, or string resource, or when a screen must work in…
android-upgrade
Workflow for changing an existing working Android project — upgrading dependencies, AGP/Gradle or SDK levels, migrating deprecated APIs, refactoring, and fixing bugs, without breaking what already works. Covers baseline verification, scoping the diff, targetSdk behaviour changes, SharedPreferences/Room data…