Security audit

Feishu File Sender Free

Security checks across malware telemetry and agentic risk

Overview

The skill is mainly for sending selected files to Feishu/Lark, but it gives unsafe credential-handling instructions and has overbroad trigger language that could cause unintended use.

Review before installing. Only use this for explicit Feishu/Lark file or image sending, confirm the recipient and file contents, and avoid regulated or secret files unless authorized. Do not paste app_secret into commands or chats; use a secret store, environment variable, or other protected credential mechanism instead.

SkillSpector

By NVIDIA

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 (6)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The skill claims coverage for unrelated functions such as file processing, format conversion, and content extraction, even though the rest of the document only implements Feishu/Lark upload-and-send flows. This overbroad scope can cause the agent to invoke the skill in unintended contexts and may justify extra tool use beyond the user’s intent, increasing the chance of unnecessary file access or data handling.

Context-Inappropriate Capability

Medium

Confidence: 98% confidence
Finding: The instructions tell the agent/user to inspect a local config file with shell commands to extract app credentials. That expands the skill from sending user-supplied files into local secret discovery, which is highly sensitive because it encourages reading unrelated local data to obtain authentication material.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger scope includes unrelated scenarios, making it more likely that the agent routes tasks to this skill when the user did not intend to send data to Feishu/Lark. In a skill with exec access and external transmission behavior, broad invocation criteria materially increase the risk of unintended data exposure.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The skill describes uploading files and recipient identifiers to Feishu/Lark APIs but does not clearly warn that local file contents and IDs will be transmitted to an external service. This reduces informed consent and can lead users to expose sensitive files or metadata without understanding the outbound data flow.

Missing User Warnings

High

Confidence: 99% confidence
Finding: The skill recommends passing app_secret directly on the command line, which can expose secrets through shell history, process listings, logs, agent transcripts, and debugging output. Because this is an authentication secret for external API access, leakage could enable unauthorized messaging or broader abuse of the Feishu application.

Missing User Warnings

High

Confidence: 99% confidence
Finding: The example embeds sensitive credentials directly into a command invocation without any caution, normalizing unsafe secret handling. In an AI-agent context, such examples are especially dangerous because they may be copied into exec calls, preserved in transcripts, and exposed to observers or logging systems.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.