write_file
Warn
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: MEDIUMNO_CODEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [NO_CODE]: The skill provides a file writing interface in SKILL.md but does not include the executable scripts (e.g., Python, JavaScript) required to verify if safety boundaries or path sanitization are enforced.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. 1. Ingestion points: The 'content' parameter. 2. Boundary markers: None defined in the metadata. 3. Capability inventory: File system write operations. 4. Sanitization: Unverifiable due to missing implementation code.
- [DATA_EXFILTRATION]: The 'path' and 'filename' parameters lack visible validation rules, creating a potential path traversal surface. Attackers could attempt to use relative paths (e.g., '../') to write to or overwrite sensitive files outside the designated 'documents' directory.
Audit Metadata