The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown documentation and instructional guidelines for reasoning techniques based on philosophical and scientific principles (Aristotle, Descartes, Feynman, Musk). No malicious code, data exfiltration patterns, or obfuscation were detected.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided problems for analytical reasoning. While this constitutes an ingestion surface for external data, the skill lacks high-risk capabilities such as network access or dynamic code execution that would facilitate a significant attack.
Ingestion points: User-supplied problem descriptions for analysis in SKILL.md.
Boundary markers: No explicit delimiters or warnings to ignore instructions within the user input are present in the guidelines.
Capability inventory: The skill configuration allows Read, Write, Edit, and Agent tools, but the skill itself does not provide scripts that invoke these tools dangerously.
Sanitization: No explicit input sanitization or filtering logic is defined in the instructions.

first-principles