directing-invisible-witness
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains no instructions aimed at bypassing AI safety protocols, overriding system prompts, or extracting internal instructions. The content is strictly educational and focused on filmmaking techniques.
- [DATA_EXFILTRATION]: There are no commands or scripts that access sensitive files, environment variables, or hardcoded credentials. No network activity or unauthorized data transfer mechanisms were found.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any external downloads, package installations, or execution of remote scripts. All content is static text providing instructional guidance.
- [COMMAND_EXECUTION]: No shell commands, subprocess calls, or system-level operations are present in either the skill definition or the test prompts.
- [OBFUSCATION]: No encoded content (such as Base64 or hex), zero-width characters, homoglyphs, or other techniques to hide malicious instructions were detected.
- [EXTERNAL_DOWNLOADS]: The skill does not reference any external URLs for downloading code, configuration, or data. References to other skills are internal identifiers and do not point to external resources.
Audit Metadata