The Agent Skills Directory

[SAFE]: The skill provides a psychological and philosophical framework designed to improve the agent's persistence during task failure. These instructions do not contain malicious commands or safety overrides.
[SAFE]: No unauthorized data access or exfiltration patterns were identified. Mentions of tools such as curl or search are framed as examples of legitimate verification steps for the agent's own work.
[SAFE]: External references point to the author's own GitHub repository (wuji-labs/nopua), which is consistent with the skill metadata and standard for documentation links.
[SAFE]: The skill does not use obfuscation, persistence mechanisms, or dynamic execution to hide intent or gain elevated permissions.

nopua-lite