Security audit

内容发布全流程（设计→落地→检查→验证→发布→沉淀）

Security checks for vulnerabilities and agentic risk

Overview

The skill is a local publishing-workflow helper whose hashing and attestation helpers are disclosed and proportionate to its release-checking purpose.

Use this as a local release-checking and attestation aid. Run the helper scripts only on directories or zip files you intend to package or verify, and treat its security and governance claims as self-attestation rather than third-party certification.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (2)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 93% confidence
Finding: The skill is presented as a documentation/publishing workflow helper, but the detected behavior extends into artifact generation, hashing, integrity verification, and attestation production. That mismatch is security-relevant because users or host systems may grant trust or invoke the skill under a lower-risk assumption, while it can influence release evidence and package integrity workflows beyond the declared purpose.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: Broad trigger keywords such as 'content', 'publish', and 'lifecycle' can cause unintended invocation in unrelated conversations. In an agent ecosystem, accidental activation increases the chance that the skill processes files, generates release artifacts, or steers workflows without the user explicitly intending to use it, which is especially concerning given the broader-than-declared behavior flagged above.

Static analysis

No suspicious patterns detected.