webhook push

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed webhook-sending helper for WeChat, DingTalk, and Feishu, with expected local key storage and outbound message delivery.

Install this only if you want your agent to send messages to configured enterprise group webhooks. Keep webhook-config.json private, use limited-purpose bot keys, and review the target group, message text, @mentions, and any image path before allowing a send, especially for sensitive or broad notifications.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (4)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 95% confidence
Finding: The skill documentation explicitly instructs reading a local user-provided configuration file containing webhook secrets and making outbound network requests, yet no permissions are declared. This creates a real security and governance gap: users and hosting platforms are not transparently informed that the skill can access local secret material and exfiltrate message content to third-party endpoints.

Description-Behavior Mismatch

Medium

Confidence: 85% confidence
Finding: The skill advertises generic webhook message pushing, but it also exposes functionality to read an arbitrary local file path and exfiltrate the file contents to an external WeCom webhook as an image. In an agent setting, this expands capability from simple outbound messaging to local file access plus network egress, which can be abused to leak sensitive local images or binary files if an attacker can influence the image_path argument.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The trigger phrases are very broad, natural-language patterns such as '通知大家明天开会' and '推送到飞书', which overlap with ordinary conversation. In an agent environment, this can cause unintended invocation of the skill and accidental transmission of user-provided content to external group webhooks without sufficiently explicit intent.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README emphasizes one-sentence message pushing to external platforms but does not clearly warn that message content will be transmitted to third-party webhook endpoints. In a conversational agent setting, this increases the risk that users or downstream systems send sensitive, private, or unintended information outside the local environment.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal