boost-prompt
Warn
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill instructs the agent to dynamically generate and execute Clojure code using the Joyride extension to write text to the system clipboard (
(vscode/env.clipboard.writeText "your-markdown-text-here")). This dynamic generation incorporates untrusted user data into an executable script block template. - [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface because it processes untrusted user prompts and embeds them directly into code execution workflows.
- Ingestion points: User-supplied task prompts are ingested during the interactive refinement workflow in
SKILL.md. - Boundary markers: Missing. There are no delimiters or instructions telling the model to isolate the untrusted user input.
- Capability inventory: The skill uses the Joyride extension to execute script code within the VS Code environment.
- Sanitization: Missing. The instructions do not specify any escaping or sanitization logic to handle special characters or double quotes within the user's input before embedding it into the Clojure code string literal.
Audit Metadata