markdown-viewer-skills
Warn
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the
markdown-viewer/skillspackage vianpx skills add. This downloads and executes content from a third-party repository that is not a recognized trusted vendor. - [COMMAND_EXECUTION]: The instructions contain shell commands for the agent to execute, including checking for file existence in the user's home directory (
~/.agents/skills/) and performing global or project-level installations usingnpx. These commands interact directly with the host environment. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to render various visualization formats (Vega-Lite, ECharts, PlantUML) from Markdown text. This functionality can be exploited if an attacker provides malicious diagram source code that the agent processes, as the complex parsers for these formats represent a significant attack surface.
- Ingestion points: Data input for Vega-Lite, ECharts, and PlantUML diagrams within Markdown files processed by the agent.
- Boundary markers: None defined to isolate the diagram code from agent instructions.
- Capability inventory: Shell execution (
npx), package management (skills add), and file system interaction (~/.agents/skills). - Sanitization: No sanitization or validation of the diagram source text is specified in the provided instructions.
Audit Metadata