tong-jincheng-interpersonal-skill
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to analyze interpersonal scenarios, which constitutes a surface for indirect prompt injection. 1. Ingestion points: Relationship questions and scenarios provided by users via triggers like '童锦程' or '深情祖师爷' and function arguments in integration examples. 2. Boundary markers: The skill instructions do not provide delimiters or negative constraints to prevent the model from following instructions embedded within the user data. 3. Capability inventory: The skill body is instructional only; provided code examples show integration with external LLM APIs (Anthropic). 4. Sanitization: No input validation or escaping mechanisms are defined in the instructions for the agent to apply to user content.
- [EXTERNAL_DOWNLOADS]: The skill documentation includes a setup command ('npx skills add hotcoffeeshake/tong-jincheng-skill') which facilitates the download of external content from a third-party source.
- [COMMAND_EXECUTION]: The README contains shell commands for platform-specific installation and includes programming examples in Node.js and Python for interacting with external services.
Audit Metadata