job-screening
Pass
Audited by Gen Agent Trust Hub on Nov 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves processing untrusted external content, which creates a potential surface for indirect prompt injection attacks. \n
- Ingestion points: Job listings ingested as text, images, URLs, or PDF/Excel files, and resume files in various formats (SKILL.md, onboarding.md). \n
- Boundary markers: No specific delimiters or instructions to ignore embedded prompts are provided when reading external content. \n
- Capability inventory: The skill executes system commands for file conversion and maintains local state in the profile directory (SKILL.md). \n
- Sanitization: Input content from job descriptions and resumes is not explicitly filtered or sanitized before being analyzed. \n- [COMMAND_EXECUTION]: The skill uses local system utilities and external Python packages to handle document processing tasks. \n
- Evidence: It invokes commands like
pdftotext,textutil, andmarkitdownvia shell execution to extract text from user-provided files (SKILL.md, onboarding.md). \n- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of a necessary third-party package for document parsing. \n - Evidence: It suggests installing
markitdown[docx,xlsx]==0.1.9into a virtual environment. This utility is a well-known tool from a recognized service provider (Microsoft) and is only installed with explicit user consent (SKILL.md).
Audit Metadata