file-reading
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous templates for executing shell commands (e.g., stat, file, pdftotext, pandoc, jq, unzip) and Python code (using pandas, openpyxl, pptx, PIL) to inspect and read uploaded files.
- [DATA_EXPOSURE]: The skill is designed to read data from the /mnt/user-data/uploads/ directory, which is the designated path for user-provided uploads on the supported platforms. This behavior is transparent and aligned with the skill's stated purpose.
- [INDIRECT_PROMPT_INJECTION]: As a file-reading utility, the skill processes untrusted external data from user uploads. While this creates an inherent surface for indirect prompt injection, the skill itself does not contain malicious instructions and promotes safer reading practices than raw file access. (1) Ingestion points: Files located at /mnt/user-data/uploads/ (identified in SKILL.md). (2) Boundary markers: The skill does not currently specify the use of delimiters (e.g., XML tags) to isolate file content from the agent's instructions. (3) Capability inventory: The skill utilizes shell command execution and Python file-reading libraries (detailed in SKILL.md). (4) Sanitization: The skill focuses on structural parsing and sampling rather than content sanitization.
Audit Metadata