CWork 工作协同

Send reports, create tasks, and manage drafts in the CWork collaboration platform with name resolution and draft confirmation workflow.

Security Scan

The skill's code and env requirements broadly match its stated purpose (CWork report/task workflows), but there are small inconsistencies and a prompt-injection detection that warrant caution before installing.

Purpose & Capability

Name/description (send reports, create tasks, manage drafts) match the included code: many report/task/contacts/analysis modules and a shared cwork-client that calls a CWork API. The declared required env CWORK_APP_KEY is appropriate for an API-key-backed integration. The API base URL in api-reference points to a specific host (sg-al-cwork-web.mediportal.com.cn), which is consistent with an enterprise integration but note the skill's source/homepage are unknown.

Instruction Scope

Runtime SKILL.md rules are scoped to the collaboration workflows (draft-first, name resolution via emp-search, LLM passed in by caller). The skill's LLM calls embed explicit systemPrompt strings in multiple files (expected for LLM-driven features). A pre-scan flagged 'system-prompt-override' in SKILL.md — likely a scanner false positive because the package legitimately supplies system prompts to LLM calls, but you should review any system/assistant prompt text to ensure it doesn't instruct unexpected behavior.

Install Mechanism

No install spec (instruction-only / code included). No external download URLs or extract steps were found. The package includes source code and a package.json but does not perform network installs during skill registration, which reduces install-time risk.

Credentials

Only CWORK_APP_KEY is required (plus optional base URL and timeout settings). This is proportionate to a single-service integration. Minor inconsistency: config/index.ts claims credentials come from an injected runtime config (setup({ apiKey })) rather than reading process.env directly, while metadata and package.json also list CWORK_APP_KEY as required — verify how you will supply the API key and how it is stored/used at runtime (in-memory vs environment vs persisted).

Persistence & Privilege

always:false and normal autonomous invocation allowed (platform default). The skill does not request system-wide config changes or other skills' credentials. No files or installers claiming to modify agent configs were observed.

Scan Findings in Context

[system-prompt-override] expected: The skill contains many LLM-invoking functions that pass explicit systemPrompt strings to an injected llmClient.generateJSON API. That pattern is expected for LLM-dependent features; the scanner likely flagged this because the SKILL.md and code include system-prompt content. Still review those prompts to ensure they do not attempt to subvert agent-level system prompts.

What to consider before installing

This package appears to implement the claimed CWork report/task features and only asks for a single API key, which is reasonable. Before installing: 1) Verify the skill's origin or contact the owner; source/homepage are missing. 2) Inspect shared/setup and runtime code to confirm how the CWORK_APP_KEY is provided and stored (prefer in-memory runtime injection, avoid persisting secrets). 3) Confirm the API base URL (sg-al-cwork-web.mediportal.com.cn) is the correct, expected service for your organization. 4) Review the LLM system prompts embedded in the code (they were flagged by a scanner); ensure they don't instruct the agent to exfiltrate data or perform actions outside the collaboration workflows. 5) If possible, test in a restricted/non-production environment and use an API key with least privilege. If you need higher assurance, ask the publisher for provenance or a signed package/source repository.

When to Use

Trigger when user wants to:

• Send, draft, or query work reports
• Create, assign, or track tasks
• Manage draft box (save/confirm/send/delete)
• Query inbox, todo list, or unread reports
• Analyze reports or summarize decisions

Quick Reference

Topic	File
Setup & security	setup.md
Standard workflows	workflow.md
All API endpoints	api-reference.md
Sub-domain skills	*/SKILL.md

Core Rules

1. Draft before send — Always save draft, show user for confirmation, then call draft-submit. Never call report-submit directly unless user explicitly says so.

2. Name resolution is built-in — Pass names directly to taskCreate and report functions. Internal emp-search runs automatically. Do not ask user for empId.

3. Ask once, act once — Collect all missing info in one message. After user confirms, call API immediately. No re-preparation loops.

4. LLM is caller-injected — Pass { llmClient } to LLM-dependent skills. This package never stores LLM credentials.

5. Output by channel — Telegram: bullets, no tables, conclusion first. Discord: tables OK. API: JSON.

Domains

• shared/ — 9 data-fetch skills (no LLM)
• reports/ — 19 skills: send, reply, draft, query, AI chat
• tasks/ — 12 skills: create, assign, track, dashboard
• decisions/ — 6 skills: summarize, extract conclusions
• closure/ — 5 skills: status check, reminder
• analysis/ — 6 skills: trends, highlights
• contacts/ — 4 skills: groups management
• llm/ — 1 skill: multi-source aggregation