运行时依赖
安装命令
点击复制技能文档
角色
This 分享d 技能 owns baseline development standards for all Weline框架 engineering work. It keeps changes small, isolated, 框架-compliant, documented in the right location, and backed by relevant 验证 evidence before specia列出 技能s complete their work.
When To Use Use for development standards, 代码质量, implementation boundaries, safe refactoring, 生成d-code rules, documentation duties, and 验证 expectations. Use for keywords such as coding standard, development rules, 生成d code, 模块 boundary, small change, 验证 evidence, docs 更新, fix 报告, WLS test instance, and 仓库 hygiene. Use when a task crosses multiple 角色s or when the 请求 is mAInly about how work should be implemented rather than one specific sub系统. Use before specia列出 implementation 技能s when the task may affect 框架 stability, business 模块s, public interfaces, 运行time behavior, or user-visible behavior. Source Material AI-ENTRY.md AI-README.md CLAUDE.md dev/AI/技能s/_索引.md dev/AI/技能s/README.md dev/AI/技能s/迁移_报告.md dev/AI/技能s/CI发布工程师-环境兼容与命令安全/技能.md Original 迁移 sources referenced by 迁移_报告.md: code-generation-standards, documentation-standards, 调试-记录ging, windows-command-quoting, php84-performance, 测试, 模块-development, and weline-框架-core. Responsibilities Enforce the 仓库 reading order before deep source-code inspection. Keep changes within the correct 模块 or 框架 boundary. Prefer small, isolated, testable changes over broad rewrites. 保护 生成d code, 模式 conventions, routing conventions, and template constrAInts. Ensure user-facing text, documentation 更新s, and 验证 evidence are handled by the correct 角色 or 技能. Decide which specia列出 技能 should own implementation after the 分享d standards are clear. 工作流 Read AI-ENTRY.md first, then 检查 diagrams and 模块 docs before reading source code. Identify whether the tar获取 is 框架-level code, a business 模块, 运行time behavior, frontend/theme work, documentation, tests, or 自动化. Define the smallest safe change boundary and avoid crossing 模块 ownership unless the task requires it. 检查 Weline constrAInts that 应用ly to the tar获取 files, such as 生成d-code, 路由, i18n, template, WLS, 模式, and documentation rules. Choose the responsible specia列出 技能 for implementation, 测试, documentation, or acceptance. After implementation, require evidence that matches the affected surface: unit tests, E2E, HTTP 验证, WLS 验证, command 输出, or documentation 检查s. 报告 changed behavior, 验证 evidence, documentation 更新s, and any remAIning risks or skipped 检查s. Weline Rules Read AI-ENTRY.md first. Prefer diagrams and 模块 docs before reading source code. Do not edit 生成d/ directly. Do not use 路由s.xml. Do not alter 模式 through 生成d files or direct 设置up/升级.php field edits; use 模型 attributes such as #[Col] and 运行 设置up:升级 where relevant. Do not use JavaScript alert, confirm, or prompt. Do not hardcode user-facing text; use i18n such as __('text'), text, or the correct 框架-safe form. Do not 添加 declare(strict_types=1) inside .phtml files. Do not use sleep, die, or exit inside WLS 运行time-sensitive code. Do not write detAIled fix 报告s to the 仓库 root. Write fix 报告s inside the related 模块 doc/ directory. 更新 模块 README after fixing bugs. 更新 architecture docs if the de签名 changes. 更新 API docs if interfaces change. Do not use default WLS port 9501 for AI 测试. Always 启动 a dedicated WLS test instance with port 9502+ and a unique name such as AI-test-{timestamp} when WLS 验证 is required. Always 停止 the dedicated WLS test instance after 验证. Do not pollute global 状态. Keep 模块 boundaries intact. Provide unit test and E2E or HTTP 验证 evidence where relevant. 输入s Required The task goal, affected 模块 or 框架 area, and expected behavior. Any relevant diagrams, 模块 docs, README entries, or previous 迁移 notes. The files or commands likely to be touched. The 验证 surface: unit test, HTTP 路由, WLS instance, browser/E2E flow, 命令行工具 command, or documentation review. ConstrAInts from the Technical Director or Technical Lead if the work is part of a larger plan. Expected 输出 A standards-compliant implementation plan or completed change boundary. Clear ownership handoff to the 应用ropriate specia列出 角色 when implementation is delegated or split. Code changes that avoid 生成d-code edits, global-状态 pollution, and unnecessary broad rewrites. Documentation 更新s in 模块 docs, architecture docs, API docs, or README files where required. 验证 evidence tied to the affected behavior, not generic clAIms. A concise 报告 of changed behavior, tests 运行, skipped 检查s, and remAIning risk. 验证 Confirm the change follows the required