Back to skill

Security audit

商业计划生成器

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward business-plan generator with disclosed OpenAI API use, optional user-directed file output, and no hidden persistence or destructive behavior.

Use quick mode for local-only template generation. If using full mode, expect your business description to be sent to OpenAI and avoid entering confidential strategy, customer data, or trade secrets. Install dependencies in a virtual environment and consider pinning the openai package version.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:51
Finding

Unpinned Third-Party Dependency Installation

Content
View full analysis
Remediation
View remediation
``` 2. Maintain dependencies in a dedicated requirements or lock file rather than relying only on documentation: ```bash python -m pip install -r requirements.txt ``` 3. Generate and verify cryptographic hashes for reproducible installation, for example with a hash-locked requirements file: ```text openai== \ --hash=sha256: ``` Install it using: ```bash python -m pip install --require-hashes -r requirements.txt ``` 4. Pin and hash all transitive dependencies through a dependency-locking tool where practical. 5. Regularly review pinned releases for known vulnerabilities and update them through a controlled testing and approval process. 6. Document use of the official Python Package Index or an authenticated internal package mirror. Avoid untrusted extra indexes. 7. Recommend installation in a dedicated virtual environment under a non-privileged account, and explicitly discourage running `pip` with administrator or root privileges. ]]>
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 (4)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill exposes capabilities to read environment variables and write files, but it does not declare any tool scope or permissions boundary in the manifest. That creates a mismatch between what a reviewer/user can see and what the skill can actually do, increasing the risk of secret access (for example API keys) or unintended filesystem writes without explicit consent.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The manifest describes a skill that generates business plans from a product or business description, but it does not mention handling credentials or accessing environment-based secrets. Reading OPENAI_API_KEY is an additional capability involving secret access, which is not obviously implied by the stated purpose itself.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

In full mode, the skill sends the user’s product/business description directly to an external LLM API without an explicit privacy notice, consent step, or redaction guidance. Business-plan inputs can contain confidential strategy, financial assumptions, customer data, or trade secrets, so silent transmission to a third party creates a real confidentiality and compliance risk.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
98% confidence
Finding

The inline help states that --lang controls output language, defaulting to auto-detection, which conveys intended behavior to users. However, args.lang is never used, and language selection is instead derived only from _detect_lang(product_desc), so the documented intent contradicts actual behavior.

Content

No source excerpt is available for this finding.