data-formulator-vscode

统一错误处理系统。在添加 API 端点、修改错误处理、添加前端 API 调用、编写错误相关测试时使用。

Greeting-triggered self-check — recognise greetings, check Edition version against the upstream tag, scan AI-Memory announcements, and report inside the greeting reply

LLM Agent 多语言注入规范。在修改 Agent 提示词、添加新的 Agent 端点、处理用户可见的后端消息（message_code）时使用。

Coordinates generic project Agency CLI, MCP authorization, M365 consent boundaries, Azure tooling, and Microsoft internal plugin adoption. Use when setting up or auditing Agency profiles, MCP servers, WorkIQ/M365 access, or curated plugins for a project.

Guides on-need adoption of an Agency MCP or Microsoft Foundry capability from the starter's capability catalog. Use when a project asks 'can we use X', 'should we adopt X', or needs to pull a specific tool/profile/Foundry capability without loading the whole surface.

Coordinates Microsoft Fabric MCP access, agents, skills, and curated marketplace plugins for a Fabric-heavy project. Use when setting up or auditing Fabric profiles, notebook-generation plugins, semantic-model review, or Fabric data-security checks.

Coordinates on-need adoption of Microsoft Foundry capabilities (IQ family, Toolboxes, Tool Search, Routines, autopilot agents, Agent Optimizer, Memory, Claude, SDKs) for a project building on Foundry Agent Service. Use when evaluating or adopting a Foundry hosted-agent capability.

Coordinates Incident Management (IcM) and DRI/on-call MCP access for a project. Use when investigating a live or recent incident, checking on-call/DRI status, or deciding whether an incident finding should become a remediation, exception, or documentation update.

服务端路径安全与文件访问编码规范。在编写文件下载路由、Agent 工具（文件读取/目录列出）、数据连接器/Loader、Workspace 路径操作、沙箱配置时使用。