The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or data exfiltration techniques were detected across the analyzed files.
[PROMPT_INJECTION]: The skill defines a specific persona and response framework but does not attempt to override agent safety filters. It includes explicit instructions to avoid promoting self-harm or misogynistic content, aligning with safety guidelines.
[DATA_EXFILTRATION]: No network operations (such as curl or wget) or access to sensitive local file paths (like .ssh or .aws) were found.
[REMOTE_CODE_EXECUTION]: The skill is purely documentation-based and does not include scripts that execute remote code or install untrusted dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill processes user input to diagnose philosophical dilemmas. While this is an ingestion point for untrusted data, the skill lacks the capabilities (e.g., file writing, network access, or command execution) required to turn an injection into a significant security threat.
[METADATA_POISONING]: YAML frontmatter and README metadata accurately reflect the skill's purpose and contain no deceptive instructions.

schopenhauer-perspective