ingest
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted content from the 'raw/' directory (Markdown, PDF, TXT) to generate wiki pages. It lacks explicit boundary markers or instructions to the agent to ignore embedded commands within these source files, creating a surface for indirect prompt injection.
- Ingestion points: Files in the 'raw/' directory (SKILL.md).
- Boundary markers: Absent.
- Capability inventory: File reading, writing, editing, and shell command execution (SKILL.md).
- Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands (mv, mkdir, test) for file archival in Step 6. While paths are shown as quoted in the templates, if the underlying agent implementation does not properly sanitize filenames or if the agent follows the template literally with a malicious filename containing shell metacharacters, it could lead to command injection.
- Evidence: bash execution for 'mkdir' and 'mv' commands in Step 6.
Audit Metadata