lf-briefing-ux
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. Ingestion points: untrusted data enters the agent context from
discovery.mdandinputs/*files during the briefing generation process. Boundary markers: there are no explicit delimiters or instructions to ignore embedded commands in the source files. Capability inventory: across all files, the agent is restricted to reading local files and writing markdown or HTML files, with no network, shell, or system-level execution capabilities. Sanitization: there is no evidence of validation or escaping for content interpolated from input files into the briefing or HTML prototype. - [COMMAND_EXECUTION]: The skill generates an
index.htmlfile containing inline JavaScript for UI prototyping. This code is used for simulating navigation and UI states (loading, error, etc.) based on data from the briefing. This generation is low risk as the code is intended for browser-based preview and is not executed within the agent's operational environment.
Audit Metadata