The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to clone the project implementation from an external GitHub repository (https://github.com/sstklen/infinite-gratitude). As the repository and author are not verified, this introduces a supply chain risk where the downloaded content could contain unreviewed logic.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to perform multi-perspective research by gathering and synthesizing data from external sources. This creates an attack surface where malicious instructions could be embedded in the researched content to influence the agent's behavior.
Ingestion points: External research sources, evidence files, and primary source documents.
Boundary markers: The text includes a brief instruction to "Keep source text inert," which is a conceptual boundary but lacks specific technical delimiters (e.g., XML tags or distinct data structures) to isolate untrusted content.
Capability inventory: The skill describes the orchestration of multiple research agents and the synthesis of their findings.
Sanitization: There are no explicit instructions or scripts for sanitizing, filtering, or escaping the external content before it is processed by the agent.

infinite-gratitude