hl-show-me
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided or session-based topics to generate visualizations, which presents a surface for indirect prompt injection.
- Ingestion points: The skill ingests data from the 'current topic' or a user-provided input string to determine what to visualize.
- Boundary markers: There are no explicit delimiters defined for the input data in the instructions to separate it from the system instructions.
- Capability inventory: The skill generates text-based diagrams and can write HTML files to the local filesystem.
- Sanitization: The instructions include a safety directive to keep HTML self-contained without remote scripts, telemetry, or data uploads, which serves as a mitigation strategy.
- [DYNAMIC_EXECUTION]: The skill dynamically generates and saves local HTML files based on user-provided data.
- Evidence: The instructions in SKILL.md direct the agent to 'write one focused HTML file', save it with a specific naming convention in a temporary directory, and 'open it in the default browser'.
- Context: While this allows for the execution of generated content, the skill explicitly forbids the inclusion of remote scripts or external assets to minimize the risk of malicious code execution.
Audit Metadata