explain-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary files from a codebase to generate an orientation page, creating a surface for indirect prompt injection. * Ingestion points: Reads project files including package.json, READMEs, source code files, and manifests to build an architecture map and trace execution flows. * Boundary markers: The skill emphasizes grounding claims in verbatim evidence with path:line citations, providing a level of verification, but does not explicitly define delimiters for untrusted content. * Capability inventory: The skill writes an HTML file to the local file system (docs/) and executes a validation script using python3. * Sanitization: Instructions require verbatim snippets but do not specify escaping logic for the resulting HTML orientation page.
- [COMMAND_EXECUTION]: The skill includes a verification step that executes a local Python script (validate_explainer.py) via the command line. This script is part of a related skill set and is used to validate the generated HTML output.
Audit Metadata