The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown-based instructions and prompt templates for persona-based reasoning. It does not define or use any tools, scripts, or binary files.
[NO_CODE]: There are no Python, Node.js, or shell scripts included in the skill. This eliminates risks associated with dynamic execution, dependency vulnerabilities, or privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided questions through a simulated expert deliberation. While this introduces a potential ingestion point for untrusted data, the skill lacks access to any tools (file system, network, or secrets), ensuring that any malicious instructions in the input cannot be executed against the host system.
Ingestion points: User-provided question following "The question is..." (SKILL.md).
Boundary markers: None explicitly enforced for the input text.
Capability inventory: None; the skill is limited to generating text responses.
Sanitization: None present in the instructions.
[DATA_EXPOSURE_AND_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file path access, or network operations was found. The instructions prioritize table formatting and logical consistency within the chat context.

collaborative-experts