skill-selector
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection (Category 8) because it is designed to autonomously infer user needs and select skills based on untrusted external data.\n
- Ingestion points: The skill analyzes user requests, open file contents, file extensions, project structure, and conversation history to determine appropriate actions.\n
- Boundary markers: Absent. The instructions encourage the agent to "proceed with confidence" and "make educated inferences" without using delimiters or verification steps to distinguish between instructions and data.\n
- Capability inventory: The skill executes a local Python script (
scripts/list_skills.py) and is empowered to recommend or trigger further skill executions based on its autonomous analysis.\n - Sanitization: None. The skill relies on LLM interpretation of file contents, which lacks formal sanitization or filtering for embedded instructions.\n- [COMMAND_EXECUTION]: The skill uses a bundled Python script (
scripts/list_skills.py) to discover available skills. This script is limited to reading metadata from local files within a specific directory and does not pose a direct threat.
Audit Metadata