The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified in the skill files.
[PROMPT_INJECTION]: The skill contains role-play instructions and an "Agentic Protocol" to guide the model's behavior as a Da Vinci persona. These instructions are functional for the skill's purpose and do not attempt to override safety constraints, bypass filters, or leak system prompts.
[EXTERNAL_DOWNLOADS]: The skill references resources from the nuwa-skills organization (e.g., nuwa-skills/davinci-skill). These are identified as legitimate vendor-owned resources and do not involve unauthorized or suspicious third-party downloads.
[DATA_EXFILTRATION]: There is no evidence of commands that access sensitive user data, environment variables, or credential files. The skill does not perform network operations to non-whitelisted domains.
[OBFUSCATION]: All content in the skill is provided in plain text. No Base64, hex encoding, zero-width characters, or hidden steganographic patterns were detected.
[COMMAND_EXECUTION]: The skill does not utilize shell commands or perform any dynamic code execution. It restricts its actions to reading its own provided reference files (research.md and demo-conversation.md).

davinci-perspective