菜单
Use when connecting an AI assistant to weclapp via the weclapp MCP server for the first time, when checking what the connection can do, or when diagnosing missing tools, permission errors, plan limits, or connection problems. Covers setup verification, a first successful read, troubleshooting, and safe escalation to support.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Use for administering the weclapp AI workspace itself — authoring and maintaining tenant workflow rules (SOPs) that steer how the assistant works, reviewing permissions, and answering governance questions from the activity history. Admin-level configuration, not business data.
Use when answering questions from live weclapp ERP data — finding records, building a customer or record 360° view, aggregating numbers, or explaining what is open, overdue, or unusual. Read-only analysis of tenant data such as customers, orders, invoices, articles, and tickets.
Use for weclapp contracts and recurring billing (Verträge) — creating or updating contracts, managing contract positions and billing intervals, checking runtimes and cancellation dates, and answering what a customer's contracts cover and cost.
Use for building weclapp CSV import files — Massenimport / Datenübernahme of articles, customers, suppliers, contacts, prices and 30+ other data types via the Import/Export-Wizard. Covers template column specs, file format rules, import order, and pre-upload validation. Not for single-record changes through the API (master-data skill) and not for auditing existing records (data-quality skill).
Use when auditing weclapp master data or records for completeness and hygiene — checking article quality, shipment readiness, finding incomplete or inconsistent records, or producing a data-quality report ("prüf meine Stammdaten"). Read-only scoring and reporting; fixes happen through the matching write skill.
Use for weclapp replenishment / disposition — deciding WHAT and HOW MUCH to reorder and turning that into draft purchase orders. Triggers include Bestellvorschlag, Nachbestellung, Disposition, Meldebestand, Reichweite, Wiederbeschaffung, "was muss ich nachbestellen", low stock, reorder proposal. Covers pulling the replenishment view and creating supplier-grouped draft purchase orders. Invoice checking and payment belong to procure-to-pay; supplier master data belongs to master-data.
基于 SOC 职业分类
| name | weclapp-get-started |
| description | Use when connecting an AI assistant to weclapp via the weclapp MCP server for the first time, when checking what the connection can do, or when diagnosing missing tools, permission errors, plan limits, or connection problems. Covers setup verification, a first successful read, troubleshooting, and safe escalation to support. |
| version | 0.1.0 |
Help the user verify their weclapp MCP connection, understand what it can do, complete a first successful read, and diagnose problems — before attempting any real work.
tenant_health_check. It confirms the tenant is reachable and reports the state of the connection.get_permissions. It returns the user's role, plan tier, enabled tool groups, and any user policy restrictions.Never promise a capability before checking it. If a tool the user asks about is not available, explain why (role, plan, or profile) instead of guessing.
Run one small, safe read so the user sees the connection working end to end:
search_entities with a familiar entity (for example customer or salesOrder), a small limit, and no writes.Work through the layers in order and report which layer blocks:
| Layer | Check | Typical fix |
|---|---|---|
| Connection | tenant_health_check fails | Credentials or tenant setup in the dashboard |
| Role | get_permissions shows viewer | Write tools are hidden by design; an admin can change the role |
| Plan | Tool group not in the plan tier | Upgrade, or use a capability the plan includes |
| User policy | Policy restricts entities or tools | A tenant admin can adjust the user policy |
| Rate limits / quotas | Requests rejected after working before | Wait for the window to reset; limits are per plan |
State clearly which layer is the cause and what the user can do. Do not retry rejected calls in a loop — limits and denials are enforced server-side and fail closed.
If a problem is not explainable by the layers above, offer escalation:
preview_escalate_to_support with a neutral, factual summary of the problem.escalate_to_support with the approval token from the preview.The same preview → approve pattern applies to preview_report_chat_for_review / write_report_chat_for_review when the user wants to flag a conversation for quality review.
Follow this contract for every change to tenant data. It applies to all write tools used by this skill and is non-negotiable.
get_permissions first.preview_* tool, present its result, and wait for the user's clear approval before calling the write tool with the approval token.This skill is for setup, verification, and diagnosis. For searching and analyzing business data, creating documents, or changing records, use the matching weclapp skill for that job once the connection works.