Back to skill

Security audit

Prompt Optimizer

Security checks for vulnerabilities and agentic risk

Overview

This is a local prompt-rewriting helper that is broad and Chinese-oriented but does not show hidden data access, persistence, network use, or destructive behavior.

Install this only if you want a Chinese-oriented prompt optimizer that may be invoked for many kinds of writing, coding, summarization, translation, and research prompts. Review custom config files before using them, and be careful with the optional batch input and output paths because the tool will read and write files you specify.

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
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 (8)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill markets itself as broadly translating vague user requests into optimized instructions, and the example trigger phrasing is very general. In agent environments, such broad activation language can cause the skill to engage on ordinary conversation or unrelated tasks, expanding its influence over prompts without clear user intent or scope boundaries.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The supported task keywords are extremely broad and overlap with common conversational requests like writing, summarizing, translating, chatting, and checking information. In a skill-routing system, this can lead to overbroad interception of user input, accidental skill invocation, and unintended prompt rewriting across many unrelated workflows.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest description says the skill is an 'AI task processor' that 'automatically translates vague requirements into precise instructions,' which is broad enough to overlap with many ordinary user requests. In a manifest file, this lacks explicit trigger scope, constraints, or negative examples to clarify when the skill should or should not activate.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill description is written entirely in Chinese, which can impose a language preference without any indication that users may choose another language. There is no documented opt-in or justification that this is a Chinese-only or region-specific skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The enhancement text explicitly instructs the model to summarize '用简洁的中文概括', which imposes Chinese as the output language. This is a natural-language locale policy issue because the file does not indicate user opt-in or provide an alternative language choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The module description, usage examples, error messages, and output text are written entirely in Chinese, and the optimization behavior also injects Chinese instructions into prompts. This indicates a fixed language/locale experience without user opt-in, which matches the policy category for language or locale constraints.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This Python test file uses Chinese natural-language prompts, labels, and expected task types throughout, indicating the skill behavior and validation are tied to a single language/locale. The policy forbids forcing a specific language without user choice, and there is no indication anywhere in the file that other languages are supported or that Chinese is optional.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The CLI help text describes the tool as an 'AI 任务处理器' (AI task processor), which suggests it can directly handle AI tasks more broadly. The implementation throughout the file only analyzes and rewrites prompt text, returning optimized prompts rather than executing or processing the underlying tasks themselves.

Content

No source excerpt is available for this finding.