email-generation
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script
scripts/sanitize-names.pyusingpython3to clean and sanitize contact data from CSV files. This involves passing file paths as arguments to the shell.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external sources (Contact CSV and Prompt Templates) to generate content. - Ingestion points: Data enters the context via the
Contact CSVand the.mdprompt template files. - Boundary markers: There are no explicit instructions or delimiters defined to prevent the AI from executing instructions that might be embedded within the CSV rows or the template body.
- Capability inventory: The skill possesses file read/write capabilities and the ability to execute shell commands (via the Python script call).
- Sanitization: While
scripts/sanitize-names.pyis used, it focuses on functional cleaning (removing titles and junk characters) rather than security sanitization to filter out malicious LLM instructions.
Audit Metadata