Security audit

协作平台卡片

Security checks for vulnerabilities and agentic risk

Overview

The skill mainly describes sending Feishu/Lark-style cards, but it gives broad automation instructions around a tool that can execute commands and send or upload content externally.

Review before installing. Use this only for explicit Feishu/Lark card sending or related image uploads, and avoid sending secrets, credentials, private screenshots, or sensitive logs unless the destination and retention are acceptable. The publisher should narrow the invocation language and document exact command and data-transfer boundaries.

A.I.G

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

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (7)

Intent-Code Divergence

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill is presented as a Feishu card sender, but the manifest and description also advertise unrelated writing, content creation, and broad automation use cases. This ambiguity can cause the agent to invoke the skill in contexts far outside its intended scope, increasing the chance of unnecessary external messaging, misuse of exec/write capabilities, or accidental data handling beyond user expectations.
Content: No source excerpt is available for this finding.

Context-Inappropriate Capability

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The usage scenarios expand the skill into intelligent analysis, data extraction, ETL pipelines, and system command automation, which are unrelated to sending collaboration-platform cards. In an agent environment, such overbroad claims can trigger the skill for high-risk tasks involving local files, automation workflows, or command execution, materially increasing attack surface and the chance of data exfiltration or unsafe actions.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The trigger conditions are overly broad and overlap with everyday requests like writing assistance and title optimization, which are not inherently tied to sending a collaboration-platform card. This can cause inappropriate auto-selection of a skill that has outbound messaging and exec/write capabilities, leading to unintended invocation and possible disclosure of user content to external services.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 84% confidence
Finding: The documentation explains inputs and capabilities but does not clearly define when the skill should or should not be invoked. In a multi-tool agent setting, unclear boundaries increase the chance that the agent routes generic content-processing tasks into a skill that can write files, execute commands, and send data externally.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The skill sends messages and uploads images to an external collaboration platform, but the documentation does not prominently warn that provided text, files, or images may leave the local environment and be stored or processed by a third party. Without an explicit exfiltration warning, users and agents may include sensitive logs, credentials, screenshots, or internal data in outbound content.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The document claims command execution is restricted to a whitelist, but no whitelist, wrapper policy, or constrained interface is actually defined, while the skill openly exposes exec. This creates a misleading safety posture that may cause operators or agents to trust execution pathways that are in fact unconstrained, increasing the risk of arbitrary command execution through downstream implementations.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 88% confidence
Finding: 文档在描述中直接写明“支持中文交互”，且未说明是否也支持其他语言或允许用户自行选择输出语言。该表述可能被实现或调用方理解为默认强制中文，从而构成语言/locale 选择不足的问题。
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.