Back to skill

Security audit

tibetan-buddhist-product-article-generator

Security checks for vulnerabilities and agentic risk

Overview

This Markdown-only skill coherently generates Tibetan Buddhist product articles and images from local drafts, with disclosed local output and expected research behavior.

Install only if you are comfortable placing drafts in the OpenClaw workspace and having product/topic details used for web verification and image prompts. Avoid unpublished or sensitive drafts unless you control network use, and be careful with --output-dir because it can place generated files outside the default output folder.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest describes only generating article files in a fixed directory, but the body documents additional outputs including PNG images, JSON metadata, extra markdown/log files, and a custom output directory option. This mismatch weakens user consent and security review because the actual filesystem write scope is broader than the declared behavior, making it easier for users or tooling to underestimate what the skill will create or where it may write.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill reads from and writes to user home-directory paths, and even creates directories/files, without clearly warning the user about local filesystem modifications. Hidden or under-disclosed file operations reduce informed consent and can cause unintended persistence, overwrites, or clutter in sensitive user-controlled locations.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documentation first presents a strict fixed output location, then later advertises a custom output directory parameter. Conflicting path guarantees are dangerous because operators may rely on the fixed-path claim while the skill can actually be redirected to write elsewhere, increasing the risk of unintended file placement or abuse.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill requires web verification based on content derived from local drafts but does not warn that local user material may be transmitted to external services or embedded in search queries. This is a meaningful privacy and data-leak risk, especially if drafts contain unpublished, personal, or proprietary information that becomes exposed through research requests.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The description requires generating a ~1000 word article in Simplified Chinese, and the generation rules reiterate Simplified Chinese as the required language. This is a language-policy constraint expressed in natural language without offering the user a choice or documenting an opt-in.

Content

No source excerpt is available for this finding.