vonneumann-perspective
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill defines a persona that processes user input to apply game theory and computational architecture frameworks. While it does not include malicious injection content, it exhibits the following characteristics related to indirect prompt injection surface area:
- Ingestion points: User queries are the primary input source for the
vonneumann-perspectiveinstructions inSKILL.md. - Boundary markers: There are no explicit delimiters (e.g., XML tags or specific markers) used to isolate user input from the skill's operational instructions.
- Capability inventory: The skill is restricted to reading its own local documentation files (
references/research.mdandexamples/demo-conversation.md). It lacks network access, file-writing capabilities, or shell execution tools. - Sanitization: No specific sanitization or filtering logic is applied to the input before it is processed by the persona framework.
- [NO_CODE]: The skill is composed entirely of Markdown files for configuration and documentation. It does not contain or attempt to execute Python, Node.js, or shell scripts.
- [SAFE]: No obfuscation, hardcoded credentials, or persistence mechanisms were found. All external references point to the author's official GitHub organization or well-known platform services.
Audit Metadata