@elvis1513/auto-coding-skill 0.1.3 → 0.1.4

package/README.md

@@ -5,6 +5,9 @@ Framework-agnostic engineering workflow skill for:
 - Claude Code
 - Codex CLI
 
+This skill is scoped to Claude/Codex workflows. During development, it should prefer already available MCP servers, installed skills, plugins, and app connectors for design, research, documentation, verification, and external system updates.
+It should also prefer multi-agent execution whenever the work can be split into parallel subtasks safely.
+
 ## Install
 
 ```bash
@@ -38,7 +41,7 @@ python3 .claude/skills/auto-coding-skill/scripts/ap.py --repo . install
 
 - `docs/ENGINEERING.md` frontmatter
 
-This frontmatter is the only manual config source (commands + deployment + docs paths).
+This frontmatter is the only manual config source (commands + local Docker runtime + docs paths).
 
 4. Start AI development by constraints:
 
@@ -50,6 +53,17 @@ This frontmatter is the only manual config source (commands + deployment + docs
 - `docs/bugs/bug-list.md`
 - `docs/tasks/summaries/**`
 
+5. Tool selection rule during execution:
+
+- Prefer current MCP/skills/plugins/apps first.
+- Fall back to shell/manual work only when those capabilities are unavailable or insufficient.
+
+6. Collaboration rule during execution:
+
+- Prefer multi-agent mode.
+- Split research, design, implementation, validation, and documentation into parallel subtasks whenever the boundaries are clear.
+- Keep one main agent responsible for integration and final gates.
+
 ## AGENTS.md Constraint Example
 
 ```md
@@ -67,16 +81,16 @@ This frontmatter is the only manual config source (commands + deployment + docs
 ### 1) docs/ENGINEERING.md
 - Purpose: single source of project config + engineering gate rules.
 - How to record:
-  - Fill YAML frontmatter once (project/commands/deployment/docs fields).
-  - Keep all environment info here only (ip/username/password/service/path/health).
+  - Fill YAML frontmatter once (project/commands/runtime/docs fields).
+  - Keep all local runtime info here only (compose file/service/container/image/health).
   - Do not duplicate config in other docs.
 
 ### 2) docs/deployment/
 - Files:
-  - `docs/deployment/deploy-runbook.md`: deployment procedure and validation checklist.
-  - `docs/deployment/deploy-records/<TASK_ID>-YYYYMMDD.md`: per-deploy execution record.
+  - `docs/deployment/deploy-runbook.md`: local Docker runtime procedure and validation checklist.
+  - `docs/deployment/deploy-records/<TASK_ID>-YYYYMMDD.md`: per-run local verification record.
 - How to record:
-  - In deploy record, write target host, service, artifact, remote path, backup, systemctl status, smoke/regression evidence.
+  - In runtime record, write compose file, service, container, image/tag, env file, docker status, smoke/regression evidence.
 
 ### 3) docs/design/
 - Files:
@@ -116,7 +130,7 @@ This frontmatter is the only manual config source (commands + deployment + docs
 - Files:
   - `docs/testing/regression-matrix.md`
 - Purpose:
-  - Full regression matrix; must be 0 FAIL.
+  - Full regression matrix against the local runtime environment; must be 0 FAIL.
 - How to record:
   - Add/maintain rows by regression ID (R-xxx), area, steps/command, expected, status, evidence.
   - If any FAIL exists, gate fails.
@@ -128,6 +142,8 @@ pip install pyyaml requests
 python3 scripts/autopipeline/ap.py run build
 python3 scripts/autopipeline/ap.py run test
 python3 scripts/autopipeline/ap.py run lint
+python3 scripts/autopipeline/ap.py run smoke
+python3 scripts/autopipeline/ap.py run regression
 python3 scripts/autopipeline/ap.py verify-api-docs
 python3 scripts/autopipeline/ap.py check-matrix
 python3 scripts/autopipeline/ap.py gen-summary T0001-1

package/cli/assets/skill/SKILL.md

@@ -1,15 +1,43 @@
 ---
 name: auto-coding-skill
-description: Use for strict project engineering workflow in Claude/Codex. Initialize docs, fill docs/ENGINEERING.md frontmatter once, then execute design->implement->gates->summary->commit/push.
+description: Use for strict project engineering workflow in Claude/Codex. Initialize docs, fill docs/ENGINEERING.md frontmatter once, then execute design->implement->local-docker-gates->summary->commit/push.
 ---
 
 # Auto Coding Skill (Claude + Codex)
 
+This skill is a Claude/Codex-only engineering workflow. During design, research, implementation, verification, and delivery, prefer already available MCP servers, installed skills, plugins, and app connectors over ad-hoc manual work whenever they can complete the task reliably.
+
+Default to multi-agent execution when the client supports it. Break work into independent design, research, implementation, validation, and documentation subtasks so Claude/Codex can run them in parallel whenever that reduces cycle time without weakening control of the main task.
+
 ## Supported clients
 
 - Claude Code
 - Codex CLI
 
+## Tooling policy
+
+Use available platform capabilities first:
+
+1) Prefer installed MCP servers for design context, documentation lookup, browser automation, issue/docs systems, and deployment/runtime inspection.
+2) Prefer already installed local skills when the task matches them.
+3) Prefer supported plugins/apps/connectors when they provide authoritative project context or can write back records.
+4) Fall back to manual shell/code workflows only when the above are unavailable, insufficient, or slower than direct execution.
+
+Typical examples:
+- Design/UI work: prefer Figma MCP and related design skills.
+- Documentation/library lookup: prefer official docs and MCP-backed doc tools.
+- Project management or knowledge base updates: prefer Linear/Notion connectors if available.
+- Browser/runtime verification: prefer Playwright/browser tools if available.
+
+## Collaboration policy
+
+Prefer multi-agent mode across the workflow:
+
+1) Split independent subtasks early when they can run in parallel.
+2) Keep the main agent on the critical path: task framing, design decisions, integration, and final quality gates.
+3) Use side agents for bounded work such as research, code slices, documentation updates, regression checks, or review passes.
+4) Do not delegate a blocking architectural decision without keeping one agent responsible for final integration and correctness.
+
 ## Entry
 
 1) Install skill files into target repo:
@@ -40,7 +68,7 @@ Fill only:
 
 This contains all manual fields:
 - `commands.*`
-- `deployment.*`
+- `runtime.*`
 - `docs.*`
 
 Do not duplicate config in other md/yaml files.
@@ -52,8 +80,9 @@ Do not duplicate config in other md/yaml files.
 3) `docs/design/**`
 4) implementation
 5) run gates via `python3 scripts/autopipeline/ap.py`
-6) update API docs + regression matrix + bug list + summary
-7) commit/push
+6) start and validate local Docker runtime
+7) update API docs + regression matrix + bug list + summary
+8) commit/push
 
 ## Commands
 

package/cli/assets/skill/data/templates/ENGINEERING.md

@@ -12,19 +12,16 @@ commands:
   smoke: ""
   regression: ""
 
-deployment:
-  host: ""
-  ssh_port: 22
-  username: ""
-  password: ""
-  service_name: ""
-  systemd_dir: "/usr/lib/systemd/system"
-  remote_app_root: ""
-  remote_jar_path: ""
-  remote_config_dir: ""
-  remote_bin_dir: ""
+runtime:
+  docker_compose_file: ""
+  docker_service: ""
+  container_name: ""
+  image: ""
+  app_port: ""
   health_base_url: ""
   health_path: ""
+  env_file: ""
+  startup_timeout_sec: 60
 
 docs:
   taskbook: "docs/tasks/taskbook.md"
@@ -41,8 +38,8 @@ docs:
 
 目标：把一次任务固化为不可跳过的流水线：  
 读任务 → 写DD → 实现 → 本地测试通过 → 静态分析+Review落盘 → 更新API Markdown+接口变更清单 →  
-部署(systemd/jar) → 重启验证 → 按API Markdown全量回归 + 回归矩阵0fail →  
-记录Bug并新增自动化回归 → 任务总结落盘 → commit → push
+本地 Docker 启动验证 → 健康检查 → 按 API Markdown 对本地环境全量回归 + 回归矩阵 0 fail →  
+记录 Bug 并新增自动化回归 → 任务总结落盘 → commit → push
 
 规则：任一步骤失败或缺产物，禁止进入下一步；未 push 视为任务未完成，禁止开始下一个任务/子任务。
 
@@ -50,7 +47,7 @@ docs:
 
 ## 0. 配置填写（必须）
 
-先填写 `docs/ENGINEERING.md` frontmatter 中的所有空值（例如 ip/用户名/密码/服务名/路径/命令）。  
+先填写 `docs/ENGINEERING.md` frontmatter 中的所有空值（例如 Docker 文件、服务名、容器名、健康检查地址、命令）。  
 禁止在其他 md/yaml 重复维护这些配置。
 
 ---
@@ -70,6 +67,36 @@ docs:
 
 ---
 
+## 1.5 工具使用策略（Claude / Codex 专属）
+
+优先使用当前环境已安装、已授权、已可访问的工具能力：
+
+1) MCP servers  
+2) 已安装 skills  
+3) plugins / apps / connectors  
+4) shell / 手工实现
+
+规则：
+- 做设计、查资料、看文档、看页面、查知识库、写回外部系统时，优先调用现有 MCP / skills / plugins / apps。
+- 能用权威工具直接完成时，不重复手写中间数据。
+- 工具不可用、无权限、结果不可靠时，才回退到本地命令或手工处理。
+- 选择工具时优先“已安装且当前项目可直接使用”的能力，而不是重新造流程。
+
+---
+
+## 1.6 多 Agent 协作策略（Claude / Codex 专属）
+
+整个流程尽可能使用多 agent 模式并行推进。
+
+规则：
+- 任务开始后，优先拆分为可并行的子任务：设计补充、资料检索、实现分块、测试验证、文档回写、review。
+- 主 agent 负责关键路径：任务定义、方案裁决、代码集成、质量门禁、最终交付。
+- 子 agent 负责边界清晰、可独立推进的工作，完成后回收结果给主 agent 集成。
+- 如果某项工作会直接阻塞主路径且难以独立定义，不要机械拆分；由主 agent 保持控制。
+- 能并行就不要串行；能拆独立 write scope 就不要让多个 agent 写同一块内容。
+
+---
+
 ## 2. Gate 流水线（强制、不可跳过）
 
 Gate-1 读任务：只从 taskbook 取范围与验收；缺信息先补 taskbook  
@@ -78,9 +105,9 @@ Gate-3 实现：严格按DD；接口变更必须同步 API Markdown
 Gate-4 本地CI：必须通过（commands.build / commands.test）  
 Gate-5 静态分析+Review：静态分析通过；docs/reviews/ 生成记录  
 Gate-6 文档：更新 api.md + 追加 api-change-log.md  
-Gate-7 部署：单机 systemd/jar；service目录固定 /usr/lib/systemd/system；失败自动回滚  
-Gate-8 重启+健康：restart后必须健康检查通过  
-Gate-9 全量回归：按 API Markdown 全量回归；回归矩阵全量 PASS（0 fail）；发现问题必须写 bug-list 并新增自动化回归用例  
+Gate-7 本地运行：必须在本地 Docker 环境启动目标服务；失败先修复再继续  
+Gate-8 健康检查：本地容器启动后必须健康检查通过  
+Gate-9 全量回归：按 API Markdown 对本地环境全量回归；回归矩阵全量 PASS（0 fail）；发现问题必须写 bug-list 并新增自动化回归用例  
 Gate-10 任务总结：必须生成 docs/tasks/summaries/<TASK_ID>.md  
 Gate-11 完成：全门禁通过后必须 commit+push
 

package/cli/assets/skill/data/templates/bridges/CLAUDE.md

@@ -1 +1,3 @@
 Follow docs/ENGINEERING.md strictly. Source of truth: docs/ENGINEERING.md.
+Prefer already available MCP servers, installed skills, plugins, and app connectors during design, research, verification, and documentation workflows.
+Prefer multi-agent mode whenever the task can be split into independent parallel subtasks without weakening integration control.

package/cli/assets/skill/data/templates/bridges/CODEX.md

@@ -1 +1,3 @@
 Follow docs/ENGINEERING.md strictly. Source of truth: docs/ENGINEERING.md.
+Prefer already available MCP servers, installed skills, plugins, and app connectors during design, research, verification, and documentation workflows.
+Prefer multi-agent mode whenever the task can be split into independent parallel subtasks without weakening integration control.

package/cli/assets/skill/data/templates/docs/deployment/deploy-records/_TEMPLATE-DEPLOY-RECORD.md

@@ -1,13 +1,14 @@
-# Deploy Record — <Task ID> — YYYY-MM-DD
+# Local Runtime Record — <Task ID> — YYYY-MM-DD
 
-- Target host:
-- Service name:
-- Artifact (local):
-- Remote jar path:
-- Updated items: jar / config / bin / service
-- Backup files (single latest): *.bak / *.tgz.bak
+- Docker compose file:
+- Docker service:
+- Container name:
+- Image/tag:
+- Exposed port:
+- Env file:
 
 ## Evidence
-- systemctl status:
+- docker ps / compose ps:
 - smoke_test:
 - api_regression:
+- health_check:

package/cli/assets/skill/data/templates/docs/deployment/deploy-runbook.md

@@ -1,12 +1,12 @@
-# Deploy Runbook（单机 systemd / jar）
+# Local Runtime Runbook（本地 Docker 验证）
 
-部署参数统一读取：`docs/ENGINEERING.md` frontmatter
-- deployment.host / deployment.ssh_port / deployment.username / deployment.password
-- deployment.service_name / deployment.systemd_dir
-- deployment.remote_app_root / deployment.remote_jar_path / deployment.remote_config_dir / deployment.remote_bin_dir
-- deployment.health_base_url / deployment.health_path
+运行参数统一读取：`docs/ENGINEERING.md` frontmatter
+- runtime.docker_compose_file / runtime.docker_service
+- runtime.container_name / runtime.image / runtime.app_port
+- runtime.health_base_url / runtime.health_path
+- runtime.env_file / runtime.startup_timeout_sec
 
-部署后必须：
+本地 Docker 启动后必须：
 - smoke-test
 - api-regression
 - 回归矩阵全量 PASS（0 fail）

package/cli/assets/skill/data/templates/docs/tasks/summaries/_TEMPLATE-TASK-SUMMARY.md

@@ -11,7 +11,7 @@
 - 目标：
 - 验收结论：PASS / FAIL（FAIL 必须说明原因与后续计划）
 
-## 2. 变更概览（代码/配置/部署）
+## 2. 变更概览（代码/配置/本地运行）
 - 关键改动点：
 - 影响模块：
 - 兼容性影响（是否破坏兼容、迁移方案）：
@@ -39,9 +39,9 @@
 - 新增/确认的 Bug（写入 `docs/bugs/bug-list.md`）：
 - 新增自动化回归用例（引用回归矩阵ID）：
 
-## 7. 部署记录（如有部署）
-- 部署记录：`docs/deployment/deploy-records/<TASK_ID>-YYYYMMDD.md`
-- systemd/service/config/bin 变更（若有）：
+## 7. 本地运行记录（如有）
+- 本地运行记录：`docs/deployment/deploy-records/<TASK_ID>-YYYYMMDD.md`
+- Docker compose/service/image/env 变更（若有）：
 
 ## 8. 风险与回滚
 - 风险：

package/cli/assets/skill/data/templates/docs/tasks/taskbook.md

@@ -2,13 +2,13 @@
 
 规则：
 1) 所有任务都写在本文件（持续续写，不另起任务文件）
-2) 允许拆子任务：每个子任务也必须走全流程（DD→实现→测试→review→接口文档→部署→回归→总结→commit→push）
+2) 允许拆子任务：每个子任务也必须走全流程（DD→实现→测试→review→接口文档→本地 Docker 验证→回归→总结→commit→push）
 3) 每个任务必须有明确验收与证据（日志/报告/文件路径）
 
 ---
 
 ## Task T0001 — <Title>
-- 状态：Planned | Designing | Implementing | Testing | Reviewing | Deploying | Done
+- 状态：Planned | Designing | Implementing | Testing | Reviewing | Runtime Verifying | Done
 - 范围（In scope）：
 - 非目标（Out of scope）：
 - 验收标准（必须可执行）：
@@ -24,7 +24,7 @@
 - API 文档：`docs/interfaces/api.md`
 - API Change Log：`docs/interfaces/api-change-log.md`
 - 本地CI：粘贴摘要或给出文件路径
-- 部署记录（如部署）：`docs/deployment/deploy-records/T0001-YYYYMMDD.md`
+- 本地运行记录：`docs/deployment/deploy-records/T0001-YYYYMMDD.md`
 - 回归矩阵：`docs/testing/regression-matrix.md`（全量PASS）
 - Bug清单（如有）：`docs/bugs/bug-list.md`
 - 任务总结（强制）：`docs/tasks/summaries/T0001.md`

package/cli/assets/skill/data/templates/docs/testing/regression-matrix.md

@@ -1,5 +1,5 @@
-# Regression Matrix（回归矩阵：必须全量 PASS，0 fail）
+# Regression Matrix（本地环境回归矩阵：必须全量 PASS，0 fail）
 
 | ID | Area | Endpoint/Feature | Test Type | Steps / Command | Expected | Status(PASS/FAIL) | Evidence |
 |---|---|---|---|---|---|---|---|
-| R-001 | API | GET /health | smoke | commands.smoke | 200 OK | PASS | <paste log path> |
+| R-001 | API | GET /health | smoke | commands.smoke | local 200 OK | PASS | <paste log path> |

package/cli/assets/skill/scripts/ap.py

@@ -92,7 +92,7 @@ def cmd_gen_summary(args: argparse.Namespace) -> None:
 - 目标：TODO
 - 验收结论：PASS / FAIL — TODO
 
-## 2. 变更概览（代码/配置/部署）
+## 2. 变更概览（代码/配置/本地运行）
 ### Git change snapshot
 - Staged files:
 {('- ' + staged.replace('\n','\n- ')) if staged else '- (none)'}

package/cli/assets/skill/scripts/core.py

@@ -104,7 +104,7 @@ def find_config(repo: Path) -> Path:
             return c
     raise APError(
         "Project config not found. Create docs/ENGINEERING.md "
-        "and put commands + deployment fields in YAML frontmatter."
+        "and put commands + runtime fields in YAML frontmatter."
     )
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@elvis1513/auto-coding-skill",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "CLI installer for auto-coding-skill (Claude Code + Codex CLI).",
   "type": "module",
   "bin": {
@@ -29,7 +29,8 @@
     "engineering",
     "workflow",
     "ci",
-    "deployment"
+    "docker",
+    "local-testing"
   ],
   "repository": {
     "type": "git",