The Agent Skills Directory

[PROMPT_INJECTION]: No evidence of malicious prompt injection or instructions to bypass safety guidelines. The adversarial tone is a functional requirement for the intended 'Cold Shower' persona and does not attempt to override the underlying model's ethical boundaries.
[DATA_EXFILTRATION]: No sensitive file access or network exfiltration patterns detected. The skill uses standard environment research tools for factual verification of market data as intended.
[REMOTE_CODE_EXECUTION]: No remote script downloads or dynamic code execution logic identified.
[COMMAND_EXECUTION]: No arbitrary shell command execution or system-level modification instructions found.
[CREDENTIALS_UNSAFE]: No hardcoded credentials or access to secret storage files such as .env or .ssh directories detected.
[SAFE]: The skill logic is confined to reasoning and critique within the agent's safe operating environment.

cold-shower