@pigcloud/skills 1.0.0

package/skills/feature-build/references/coding-checklist.md

@@ -0,0 +1,97 @@
+# 编码检查清单
+
+> 以规则层为主，保留可执行检查，不重复展开规则正文。默认基线见 [rules/overlays/pig-cloud.md](/D:/work/java/pig-skills/rules/overlays/pig-cloud.md)。
+
+---
+
+## Entity Layer
+
+- 基类继承是否符合 [rules/overlays/pig-cloud/entity.md](/D:/work/java/pig-skills/rules/overlays/pig-cloud/entity.md)
+- 主键策略是否符合项目约定
+- 状态字段是否使用约定枚举编码
+- 表名是否显式且稳定
+
+---
+
+## DTO Layer
+
+- CreateDTO 是否具备必要校验
+- UpdateDTO 是否包含 id
+- VO 是否做好脱敏和输出边界控制
+- DTO / VO / Entity 是否严格分离
+
+---
+
+## Mapper Layer
+
+- 是否只负责数据访问
+- 是否保持 lambda query / update 风格
+- 是否避免字符串列名和 `${}` 注入风险
+- 是否避免业务逻辑下沉到 Mapper
+
+---
+
+## Service Layer
+
+- 是否通过 Service 统一组织业务编排
+- 是否在 Service 层管理事务边界
+- 是否避免事务内远程调用
+- 是否避免把接口层语义带入业务层
+
+---
+
+## Controller Layer
+
+- 是否只负责接入、参数接收和结果返回
+- 是否通过 Service 调用业务逻辑
+- 是否符合 [rules/overlays/pig-cloud/validation.md](/D:/work/java/pig-skills/rules/overlays/pig-cloud/validation.md)
+- 是否避免直接访问 Mapper 和写事务
+
+---
+
+## Comment Layer
+
+- 类注释是否一句话说明职责
+- 方法注释是否说明目的、输入、输出和异常
+- 复杂逻辑是否有必要的步骤说明
+- 外部调用和状态流转是否留下注释
+
+---
+
+## Logging Layer
+
+- 外部调用是否记录接口名、参数和耗时
+- 业务事件是否记录关键上下文
+- 异常日志是否保留足够上下文
+- 敏感数据是否已脱敏
+
+---
+
+## Shared Tools
+
+- String 是否使用统一工具
+- JSON 是否使用统一工具
+- Bean Copy 是否使用统一工具
+- DateTime 是否使用统一工具
+- Password 是否使用统一工具
+- Pagination 是否使用统一工具
+- Exception 是否使用统一异常类型
+
+---
+
+## Anti-rationalization
+
+- 是否出现“临时方案”“以后再优化”“不会发生”的借口
+- 是否把边界问题推迟到未来处理
+- 是否把细节复杂度低估
+- 是否把覆盖率、性能、异常和安全问题合理化
+
+---
+
+## Verification Gates
+
+- 编译是否通过
+- 每个任务的本地自测是否有证据
+- 规则检查是否通过
+- 分层检查是否通过
+- 安全检查是否通过

package/skills/feature-build/references/comment-specification.md

@@ -0,0 +1,102 @@
+# 娉ㄩ噴瑙勮寖
+
+> 浠ｇ爜娉ㄩ噴鏍囧噯锛岀敤浜庝繚鎸佸彲璇绘€у拰鍙淮鎶ゆ€с€傝繖閲屽彧淇濈暀娉ㄩ噴瀵嗗害銆佹ā鏉垮拰妫€鏌ラ」锛涘叿浣撲笟鍔?鍒嗗眰鍩虹嚎浼樺厛寮曠敤 [rules/overlays/pig-cloud.md](/D:/work/java/pig-skills/rules/overlays/pig-cloud.md) 鍙婂叾瀛愯鍒欍€?
+## Class Comment
+
+### 蹇呰鍐呭
+
+| 鍐呭 | 璇存槑 |
+|------|------|
+| 绫昏鏄?| 涓€鍙ヨ瘽鎻忚堪鑱岃矗 |
+| 浣滆€?| 鍒涘缓鑰?|
+| 鏃ユ湡 | 鍒涘缓鏃ユ湡 |
+| 鐗堟湰 | 鍙€?|
+
+### 妯℃澘
+
+```java
+/**
+ * 璁㈠崟鏈嶅姟瀹炵幇绫伙紝澶勭悊璁㈠崟鐩稿叧涓氬姟閫昏緫
+ *
+ * @author zhangsan
+ * @date 2026-06-16
+ */
+@Service
+@RequiredArgsConstructor
+public class OrderServiceImpl implements OrderService {
+}
+```
+
+## Method Comment
+
+### 蹇呰鍐呭
+
+| 鍐呭 | 璇存槑 |
+|------|------|
+| 鏂规硶璇存槑 | 涓€鍙ヨ瘽鎻忚堪鍔熻兘 |
+| 鍙傛暟璇存槑 | 姣忎釜鍙傛暟鐨勫惈涔?|
+| 杩斿洖璇存槑 | 杩斿洖鍊煎惈涔?|
+| 寮傚父璇存槑 | 鍙兘鎶涘嚭鐨勫紓甯?|
+| 鍓嶇疆鏉′欢 | 鎵ц鍓嶉渶瑕佹弧瓒崇殑鏉′欢 |
+| 鍚庣疆鏉′欢 | 鎵ц鍚庣殑鐘舵€佸彉鍖?|
+
+### 妯℃澘
+
+```java
+/**
+ * 鍒涘缓璁㈠崟锛屾牎楠屽簱瀛樺苟鐢熸垚璁㈠崟璁板綍
+ *
+ * @param dto 璁㈠崟鍒涘缓璇锋眰DTO
+ * @return 璁㈠崟ID
+ * @throws BusinessException 搴撳瓨涓嶈冻鏃舵姏鍑?
+ */
+public Result<Long> create(OrderCreateDTO dto) {
+}
+```
+
+## Logic Comment
+
+### 閫傜敤鍦烘櫙
+
+- 澶氭潯浠跺垽鏂?- 鐘舵€佹祦杞?- 澶嶆潅璁＄畻
+- 澶栭儴璋冪敤
+- 鏁版嵁杞崲
+- 鍒嗛〉/鎵归噺閫昏緫
+
+### 瑙勫垯
+
+- 鐢?`Step 1/2/3` 鏍囪娴佺▼
+- 鍙В閲婁笟鍔℃剰鍥撅紝涓嶈В閲婃樉鑰屾槗瑙佺殑璇硶
+- 澶嶆潅閫昏緫蹇呴』鐣欎笅鑳界湅鎳傜殑涓婁笅鏂?
+## External Call Comment
+
+### 蹇呰鍐呭
+
+- 璋冪敤鐩殑
+- 璋冪敤鎺ュ彛
+- 浼犲叆鍙傛暟
+- 棰勬湡杩斿洖
+- 寮傚父澶勭悊
+
+### 绀轰緥
+
+```java
+// 杩滅▼璋冪敤锛氭煡璇㈢敤鎴蜂俊鎭紝鐢ㄤ簬璁㈠崟鍒涘缓鍓嶆牎楠?UserDTO user = userApi.findById(dto.getUserId()).assertSucceed();
+```
+
+## Comment Density
+
+| 灞傜骇 | 寤鸿瀵嗗害 |
+|------|---------|
+| Entity | 绫绘敞閲?+ 瀛楁娉ㄩ噴 |
+| DTO/VO | 绫绘敞閲?+ 瀛楁娉ㄩ噴 |
+| Mapper | 绫绘敞閲婂嵆鍙?|
+| Service | 绫绘敞閲?+ 鏂规硶娉ㄩ噴 + 閫昏緫娉ㄩ噴 |
+| Controller | 绫绘敞閲?+ 鏂规硶娉ㄩ噴 |
+| Convert | 绫绘敞閲婂嵆鍙?|
+
+## Anti-rationalization
+
+- 浠ｇ爜寰堟竻妤氾紝涓嶉渶瑕佹敞閲?- 娉ㄩ噴浼氳繃鏃讹紝鎵€浠ヤ笉鍐?- 浠ュ悗鍐嶈ˉ娉ㄩ噴
+- IDE 鍙互鑷姩鐢熸垚灏卞浜?
+

package/skills/knowledge-capture/SKILL.md

@@ -0,0 +1,84 @@
+---
+name: knowledge-capture
+description: Capture reusable decisions, requirements knowledge, realtime code facts, codewiki facts, and handoff notes when downstream work needs a durable summary
+lifecycle_stage: capture
+rule_profile: capture
+dependencies: []
+triggers:
+  - capture
+  - handoff
+  - note
+  - summary
+  - decision log
+  - knowledge
+inputs:
+  - approved artifacts
+  - findings
+  - decisions
+outputs:
+  - reusable note
+  - knowledge page
+  - handoff summary
+workflow:
+  - collect the produced artifact
+  - normalize it into a reusable note
+  - separate requirement knowledge, realtime code facts, CodeWiki facts, and decision-log content
+  - link it to the originating skill
+  - publish the handoff summary for downstream reuse
+gates:
+  - stop at capture and organization
+  - do not redo upstream analysis
+  - do not generate new requirements or code
+refs:
+  - skills/references/skill-reference-matrix.md
+  - skills/references/rule-loading-map.md
+  - skills/references/flow-test-cases.md
+  - skills/references/full-chain-replay-scenarios.md
+  - skills/references/prompt-replay-checklist.md
+  - docs/codewiki-knowledge-base.md
+---
+
+# Knowledge Capture
+
+## Purpose
+
+Store decisions, artifacts, and handoff notes so other skills can reuse them later.
+
+## Suitable / Unsuitable
+
+- Suitable: summaries, decision logs, reusable knowledge, artifact handoff, CodeWiki facts, requirements knowledge, realtime code facts
+- Unsuitable: requirement discovery, design, implementation, review, test authoring
+
+## Workflow
+
+1. Collect the produced artifact.
+2. Normalize it into a reusable note.
+3. Separate requirement knowledge, realtime code facts, CodeWiki facts, decision log entries, and pitfall notes.
+4. Link it back to the originating skill.
+5. Publish the handoff summary for downstream reuse.
+
+## Replay Signals
+
+- Input signal: approved artifacts, findings, or decisions.
+- Output to verify: reusable note, knowledge page, handoff summary, decision log entry, codewiki update, enterprise knowledge note.
+- Stop signal: redoing upstream analysis or generating new requirements/code.
+- Handoff signal: the reusable note is ready for later reuse by other skills.
+
+## Stop Rules
+
+- Stop at capture and organization.
+- Do not redo upstream analysis.
+- Do not generate new requirements or code.
+- Do not merge requirement knowledge, code facts, and decisions into one blob.
+
+## Examples
+
+- Input: "Capture the decisions and pitfalls from this task."
+- Output: reusable note, knowledge page, handoff summary
+- Stop: do not re-open the design or implementation debate
+
+## References
+
+- `skills/references/skill-reference-matrix.md`
+- `skills/references/rule-loading-map.md`
+- `skills/references/flow-test-cases.md`

package/skills/performance-check/SKILL.md

@@ -0,0 +1,117 @@
+---
+name: performance-check
+description: Review a change for latency, throughput, or resource bottlenecks when performance findings are needed
+lifecycle_stage: review
+rule_profile: performance
+dependencies:
+  - feature-build
+triggers:
+  - performance
+  - slow
+  - latency
+  - cache
+  - bottleneck
+  - slow path
+  - hotspot
+inputs:
+  - diff
+  - hotspots
+  - response timings
+outputs:
+  - performance findings
+  - bottleneck map
+  - optimization advice
+  - measured path summary
+  - review summary
+workflow:
+  - locate the expensive path
+  - inspect queries, cache, memory, and concurrency
+  - route correctness-only issues to `code-review`
+  - rank the bottlenecks by impact
+  - produce optimization advice
+gates:
+  - stop at performance findings
+  - do not rewrite code
+  - do not become a general review
+refs:
+  - skills/references/rule-loading-map.md
+  - skills/references/prompt-replay-checklist.md
+  - skills/references/full-chain-replay-scenarios.md
+  - rules/index.md
+---
+
+# Performance Check
+
+## Purpose
+
+Find performance bottlenecks and optimization opportunities.
+
+## Suitable / Unsuitable
+
+- Suitable: slow paths, latency, cache behavior, query cost, concurrency, measurable bottlenecks
+- Unsuitable: implementation work, broad code review, requirement discovery, generic correctness review
+
+## Quick Start
+
+- Identify the hottest path first.
+- Check queries, cache, memory, and concurrency in that order.
+- Return a bottleneck map plus optimization advice.
+- If the issue is only about correctness or maintainability, hand off to `code-review`.
+
+## Report Format
+
+- Use one entry per bottleneck.
+- Include `location`, `bottleneck type`, `evidence`, `expected impact`, and `optimization advice`.
+- Anchor each bottleneck to the smallest relevant performance surface:
+  - query, cache path, allocation path, concurrency path, or hot loop
+- Keep the bottleneck map explicit about CPU, memory, network, database, or allocation pressure.
+- Keep the review summary short when no performance issue is present.
+- If there are no findings, explain what was measured or inspected and why it was acceptable.
+
+## Inputs / Outputs
+
+- Inputs:
+  - diff
+  - hotspots
+  - response timings
+- Outputs:
+  - performance findings
+  - bottleneck map
+  - optimization advice
+  - measured path summary
+
+## Workflow
+
+1. Locate the expensive path.
+2. Inspect queries, cache, memory, and concurrency.
+3. Route correctness-only issues to `code-review`.
+4. Rank the bottlenecks by impact.
+5. Produce optimization advice.
+
+## Replay Signals
+
+- Input signal: diff, hotspots, response timings, or a performance concern.
+- Output to verify: performance findings, bottleneck map, optimization advice, measured path summary, review summary.
+- Stop signal: rewrite code or turn into a general review.
+- Handoff signal: the findings are ready for `knowledge-capture`.
+
+## Gotchas
+
+- Do not stop at vague "slow" labels.
+- Do not turn the task into a generic review.
+- Do not skip the bottleneck map.
+- Do not give advice without an identifiable path.
+- Do not file a generic correctness issue as a performance finding.
+
+## Stop Rules
+
+- Stop at performance findings.
+- Do not rewrite code.
+- Do not become a general review.
+
+## References
+
+- `skills/references/rule-loading-map.md`
+- `skills/references/prompt-replay-checklist.md`
+- `skills/references/full-chain-replay-scenarios.md`
+- `rules/index.md`

package/skills/product-intake/SKILL.md

@@ -0,0 +1,98 @@
+---
+name: product-intake
+description: Capture product intent and separate it from project facts, prototypes, and current implementation facts into a PRD-ready brief when the goal, scope, or priority is still fuzzy
+lifecycle_stage: intake
+rule_profile: intake
+dependencies: []
+triggers:
+  - business goal
+  - user need
+  - PRD
+  - scope
+  - priority
+  - problem framing
+inputs:
+  - business goals
+  - user feedback
+  - meeting notes
+  - release context
+  - prototype notes
+  - current implementation facts
+outputs:
+  - brief
+  - scope
+  - acceptance outline
+  - current project facts
+  - prototype notes
+  - implementation facts
+  - open questions
+workflow:
+  - confirm the problem and desired outcome
+  - separate in-scope and out-of-scope
+  - capture current project facts, prototypes, and implementation facts separately from product intent
+  - capture acceptance criteria and open questions
+  - hand off a PRD-ready brief plus project facts, prototype notes, implementation facts, and open questions to spec-refinement
+gates:
+  - stop at product-level framing
+  - do not write implementation details
+  - do not choose rule bundles
+  - do not produce code-oriented specs
+refs:
+  - skills/references/requirements-separation-map.md
+  - skills/references/project-requirement-alignment.md
+  - skills/references/rule-loading-map.md
+  - skills/references/flow-test-cases.md
+  - skills/references/full-chain-replay-scenarios.md
+  - skills/references/prompt-replay-checklist.md
+  - skills/spec-refinement/SKILL.md
+---
+
+# Product Intake
+
+## Purpose
+
+Turn business input into a PRD-ready brief. This skill ends when the product intent is clear enough for engineering refinement.
+
+## Suitable / Unsuitable
+
+- Suitable: goals, user pain points, scope, priority, release intent, meeting notes, product context, prototypes, current implementation facts
+- Unsuitable: implementation details, rule choice, coding tasks, test design, code review
+
+## Workflow
+
+1. Confirm the problem, target user, and expected outcome.
+2. Separate in-scope, out-of-scope, and assumptions.
+3. Capture current project facts, prototypes, and implementation facts separately from product intent.
+4. Write acceptance criteria and open questions.
+5. Hand off a PRD-ready brief plus project facts, prototypes, and implementation facts to `spec-refinement`.
+6. Stop before implementation analysis or rule selection.
+
+## Replay Signals
+
+- Input signal: business goal, user pain point, release intent, or meeting notes.
+- Output to verify: product objective, scope, acceptance outline, current project facts, prototype notes, implementation facts, open questions, assumptions.
+- Stop signal: any implementation detail, architecture choice, or rule.
+- Handoff signal: a PRD-ready brief is ready for `spec-refinement`.
+
+## Examples
+
+- Input: "We need a login reminder feature for inactive users."
+- Output: PRD-ready brief, scope, acceptance outline, current project facts, open questions
+- Stop: do not decide implementation details or rule bundles
+
+## Stop Rules
+
+- Stop at product-level framing.
+- Do not write implementation details.
+- Do not write technical design.
+- Do not choose rule bundles.
+- Do not invent technical constraints.
+- Do not generate code-level specs.
+- Do not produce code-oriented specs.
+
+## References
+
+- `skills/references/requirements-separation-map.md`
+- `skills/references/rule-loading-map.md`
+- `skills/references/flow-test-cases.md`
+- `skills/spec-refinement/SKILL.md`

package/skills/project-bootstrap/SKILL.md

@@ -0,0 +1,80 @@
+---
+name: project-bootstrap
+description: Bootstrap a new workspace with the right skills and rules bundle when a project needs initial setup
+lifecycle_stage: scaffold
+rule_profile: scaffold
+dependencies: []
+triggers:
+  - bootstrap
+  - init
+  - new project
+  - workspace setup
+  - starter
+  - setup
+inputs:
+  - client type
+  - workspace markers
+  - desired rules bundle
+outputs:
+  - starter structure
+  - installed bundle
+  - setup notes
+workflow:
+  - detect the target client
+  - install the required skill and rules layout
+  - write the starter files
+  - hand off to the router or intake skill
+gates:
+  - stop after setup
+  - do not design product scope
+  - do not generate application code
+refs:
+  - skills/references/rule-loading-map.md
+  - skills/references/prompt-replay-checklist.md
+  - skills/references/full-chain-replay-scenarios.md
+  - rules/index.md
+---
+
+# Project Bootstrap
+
+## Purpose
+
+Create a new workspace with the right skills and rules layout.
+
+## Suitable / Unsuitable
+
+- Suitable: workspace setup, starter files, rules bundle installation
+- Unsuitable: product discovery, architecture design, feature implementation
+
+## Workflow
+
+1. Detect the target client.
+2. Install the required skill and rules layout.
+3. Write the starter files.
+4. Hand off to the router or intake skill.
+
+## Replay Signals
+
+- Input signal: client type, workspace markers, desired rules bundle.
+- Output to verify: starter structure, installed bundle, setup notes.
+- Stop signal: product scope design or application code generation.
+- Handoff signal: the workspace is ready for `workflow-router` or `product-intake`.
+
+## Examples
+
+- Input: "Set up a new workspace for Codex with the right skills and rules."
+- Output: starter structure, installed bundle, setup notes
+- Stop: do not design product scope or generate application code
+
+## Stop Rules
+
+- Stop after setup.
+- Do not design product scope.
+- Do not generate application code.
+
+## References
+
+- `skills/references/rule-loading-map.md`
+- `skills/references/prompt-replay-checklist.md`
+- `skills/references/full-chain-replay-scenarios.md`
+- `rules/index.md`

package/skills/references/agent-personas.md

@@ -0,0 +1,34 @@
+# 角色画像
+
+这个文档定义的是面向审查的角色，以及它们对应的 canonical skill。
+
+## 角色映射
+
+| 角色 | 主要 skill | 审查重点 | 输出 |
+|---|---|---|---|
+| 代码审查者 | `code-review` | 分层、命名、可维护性 | 审查报告 |
+| 安全审查者 | `security-review` | 注入、密钥、授权 | 安全报告 |
+| 性能审查者 | `performance-check` | N+1、缓存、延迟、内存 | 性能报告 |
+| 架构审查者 | `spec-refinement` | 边界检查、模块拆分、依赖方向 | 设计补充 |
+| 测试审查者 | `test-design` | 覆盖率、边界场景、回归 | 测试报告 |
+
+## 触发关键词
+
+- `code-review`：review、diff、refactor、maintainability
+- `security-review`：security、injection、secret、permission
+- `performance-check`：performance、slow、cache、latency
+- `spec-refinement`：design、architecture、boundary、module split
+- `test-design`：test、coverage、regression、boundary case
+
+## 协作链路
+
+```text
+code-review -> security-review -> performance-check
+spec-refinement -> technical-design -> feature-build
+test-design -> code-review
+```
+
+## 输出规则
+
+- 每个角色只产出一个聚焦的工件。
+- 只有相关审查门禁都通过后，才合并结果。