Back to skill

Security audit

AI智能写作助手Pro

Security checks for vulnerabilities and agentic risk

Overview

This Chinese AI-writing skill is purpose-aligned, but it needs Review because its search feature can execute shell commands from a user-supplied topic and runs an unverified local search script with inherited environment secrets.

Review before installing. Use only with trusted topic input, avoid setting TAVILY_API_KEY until the shell execution is fixed, and do not rely on the advertised rewrite/SEO/dedup features unless those scripts are added. The publisher should replace execSync string execution with argument-array process execution or a direct Tavily API client, pass only the required environment variables, add clear privacy and overwrite warnings, and verify or remove the external local-skill dependency.

Vulnerability Patterns
  • Tool Hijacking and SpoofingModifies or replaces tools so legitimate-looking calls execute attacker logic
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/write.mjs:159
Finding

OS Command Injection Through the User-Controlled Topic Argument

Content
View full analysis
Remediation
View remediation

T07 · Tool Hijacking and Spoofing

Warning
Location
scripts/write.mjs:156
Finding

Execution of an Unverified External Skill With Inherited Environment Secrets

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (10)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The script constructs a shell command and executes another script using execSync, which is a dangerous capability expansion for a writing tool. Because the command embeds user-controlled topic text inside a shell string, this introduces command injection risk in addition to cross-skill execution, potentially allowing arbitrary command execution under the current user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The natural-language instructions and descriptions are entirely in Chinese, which effectively forces a specific language for users without opt-in or explanation. Under the policy, language constraints should either offer user choice or be clearly documented and justified as region-specific.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This markdown file advertises '自动搜索相关热点话题' using Tavily AI Search, which implies outbound network requests based on user-provided topics. The README does not warn users that their prompts/topics may be sent to a third-party search service, creating a missing disclosure about privacy-affecting behavior.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill declares an environment dependency (TAVILY_API_KEY) but does not define any explicit tool scope such as permissions or allowed-tools. That creates ambiguity about what runtime actions the skill may perform with the secret and weakens least-privilege controls, increasing the risk of unintended network use or secret exposure if supporting scripts access the environment unsafely.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest advertises a broader AI writing suite, including 查重, SEO 优化, and 改写润色. The code in this file implements style-based article generation and appends search results, but contains no logic for duplicate checking, SEO analysis/optimization, or rewriting existing text, so the described behavior does not match the actual implementation here.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The script reads an external API key from the environment and conditionally enables network-backed behavior via a separate search skill, expanding capability beyond simple local content templating. In a writing assistant, this creates unexpected data-flow and dependency risk: user topics can be sent to external services, and behavior now depends on another skill outside this file’s trust boundary.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

User-controlled topic input is inserted directly into a shell command without safe argument separation, making this more than a disclosure issue: it is an exploitable command injection sink. In this skill context, topics are expected to be arbitrary free-form text, which increases the likelihood that special shell metacharacters could be supplied intentionally or accidentally.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

The usage example documents saving output to '~/article.md', which affects local user data by writing to the filesystem. The README does not include any caution about file creation or possible overwrite behavior, so users are not warned about this data-affecting operation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The script writes generated content to a user-specified file path when --output is provided. While it logs after saving, the help and surrounding comments do not warn that this option will overwrite or create files on disk, which is a safety-relevant side effect.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The generated timestamp is forced to use the zh-CN locale and Asia/Shanghai timezone, which imposes a specific language/locale behavior without user opt-in. The file does not indicate that this is a region-specific tool or provide a configurable locale setting.

Content

No source excerpt is available for this finding.