iso-standards-readiness
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze user-provided JSON and Markdown files. While the Python scripts implement rigorous structural validation, nesting limits, and path traversal blocks, the contents of these external files influence the agent's context. This ingestion process is an inherent surface for indirect prompt injection.
- [COMMAND_EXECUTION]: Although the skill manifest requests the
Bashtool, the included Python scripts do not execute shell commands or spawn subprocesses. All logic is implemented using the Python standard library. - [REMOTE_CODE_EXECUTION]: There is no evidence of remote code execution. The scripts do not perform network operations, download external scripts, or use dynamic evaluation functions like
eval()orexec(). - [DATA_EXFILTRATION]: The skill operates in a bounded local environment. Access to the file system is restricted through relative path validation and symlink rejection, and the lack of network capabilities prevents the transmission of data to external servers.
Audit Metadata