The Agent Skills Directory

[SAFE]: The skill serves as an instructional framework for selecting the appropriate search or research tool based on user intent. It includes explicit constraints to limit result sizes (max_results: 3) and prevent context bloat, demonstrating a focus on performance and cost efficiency.
[SAFE]: No executable code, shell scripts, or external dependencies are included. The skill operates purely through high-level instructions to the agent on how to call existing MCP (Model Context Protocol) tools.
[SAFE]: There are no signs of prompt injection, data exfiltration, or persistence mechanisms. The tool selection chain encourages the use of specialized, local tools (like Nx or Graphite) before reaching out to external search services, which is a security best practice for data locality.

perplexity