Security audit

詹明明·选题先试水

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed local workflow for drafting and tracking X post tests, with purpose-aligned file and memory updates but no evidence of deception, exfiltration, or unsafe execution.

Install this only if you are comfortable with a skill reading your local ZMM vault materials and X-post history, then saving drafts, updating pipeline records, using the clipboard, moving published drafts, and writing workflow memory. It does not post to X for you, and the inspected artifact does not show network exfiltration or destructive behavior.

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

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 (3)

Description-Behavior Mismatch

Medium

Confidence: 87% confidence
Finding: The skill goes beyond content generation and instructs persistent local modifications: saving drafts, updating pipeline files, moving published items, and writing memory records. Without explicit confirmation gates and clear disclosure, a user invoking a writing helper could unintentionally alter local state, causing integrity issues in notes, workflows, or records.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: Broad natural-language triggers like 'write a short post to validate' or 'this idea test the waters' can match ordinary conversation and cause accidental invocation. In a skill that can read local files and modify persistent records, ambiguous activation increases the risk of unintended side effects from benign chat input.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The workflow directs writing drafts, modifying pipeline records, moving files, and updating memory without an explicit user-facing warning that local data will change. Hidden or insufficiently disclosed state changes are dangerous because they can overwrite, reorganize, or pollute user data under the guise of a text-generation task.

Static analysis

No suspicious patterns detected.