guizang-ppt

Warn

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill is a stub designed to download and install its full implementation from a remote repository located at github.com/op7418/guizang-ppt-skill.
  • [COMMAND_EXECUTION]: The instructions include shell commands utilizing npx skills add with the -y flag to automate the installation of third-party code. This bypasses user confirmation prompts during the installation of remote scripts.
  • [PROMPT_INJECTION]: The skill processes untrusted user-supplied data, such as raw text and content outlines, to generate presentations. This introduces a surface for indirect prompt injection where instructions embedded in the input data could manipulate the agent's behavior during the PowerPoint generation process.
  • Ingestion points: Input content, outlines, or raw text processed by the guizang-ppt command as described in SKILL.md.
  • Boundary markers: No delimiters or protective instructions (e.g., "ignore embedded instructions") are present in the provided files.
  • Capability inventory: The skill is designed to generate and save .pptx files to the local file system.
  • Sanitization: No evidence of input validation, escaping, or sanitization is provided in the installer stub.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 25, 2026, 06:33 PM
Security Audit — agent-trust-hub — guizang-ppt