Back to skill

Security audit

企业微信智能办公助手

Security checks for vulnerabilities and agentic risk

Overview

The skill is not deceptive, but it should be reviewed because it can automatically gather sensitive workplace data and send the report to chat without clear recipient or approval controls.

Install only in an organization-approved workspace. Configure least-privilege WeCom and smart-sheet access, disable or gate automatic sending until recipients are explicitly chosen, and avoid scheduled delivery of confidential schedules, todos, or team metrics without consent and retention guidance.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The scheduled execution example instructs the system to generate a daily report from WeCom data and send it to Feishu, creating cross-system transfer of potentially sensitive corporate information. Because the README provides no warning, approval requirement, recipient restrictions, or sanitization guidance, this could lead to unintentional disclosure of internal schedules, meeting details, to-dos, and KPI data into another platform.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill is designed to aggregate schedules, meetings, to-dos, and team spreadsheet data into a single report, which materially increases the exposure of sensitive workplace information. Without clear privacy warnings, consent expectations, access scoping, or handling guidance, users may deploy it in ways that over-collect or over-share internal business data.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documentation for the scheduled execution explicitly says the skill will 'generate today's morning brief and send it to Feishu' at L104. However, the earlier architecture and workflow sections only describe querying WeCom-related sources and rendering a PNG card, with no Feishu integration or send action documented anywhere, creating an intent/documentation contradiction.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill describes automatic daily aggregation of schedules, meetings, to-dos, and team spreadsheet data, plus sending the rendered report to Feishu/WeCom, but does not clearly warn users that sensitive workplace data will be collected, combined, and shared automatically. Aggregation across systems materially increases sensitivity, and silent outbound delivery to group channels can expose confidential operational information to unintended recipients.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manual trigger phrases are broad, natural-language requests like '今天的日程是什么' and '帮我看看待办', which can easily match ordinary conversation and invoke the skill unexpectedly. In this skill's context, unintended activation is more dangerous because execution aggregates multiple sources of workplace data and may generate/share a consolidated report, increasing the chance of over-collection or disclosure beyond the user's immediate intent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manual trigger phrase "今天的日程" is broad and plausibly overlaps with ordinary user conversation, which can cause the workflow to run unintentionally. In this skill, accidental invocation is more sensitive because execution aggregates calendar, meeting, to-do, and team metrics data and may deliver a rendered report through corporate messaging channels.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

The manifest hard-codes the schedule timezone to Asia/Shanghai, which is a locale-specific setting. The file does not indicate that users can choose their locale/timezone or that the skill is intentionally restricted to a China-specific deployment context.

Content

No source excerpt is available for this finding.