The Agent Skills Directory

[SAFE]: The skill provides a legitimate interface to the public grep.app search service. It uses standard MCP (Model Context Protocol) client libraries to communicate with a hosted search provider.
[COMMAND_EXECUTION]: The skill executes a local Python script scripts/mcpcall.py using uv run. This script is included in the skill and its behavior is limited to making network requests to https://mcp.grep.app and printing the search results. No arbitrary shell execution or dangerous command construction was detected.
[EXTERNAL_DOWNLOADS]: The skill communicates with https://mcp.grep.app, which is the intended functional endpoint for the search service. This is documented and consistent with the skill's purpose.
[INDIRECT_PROMPT_INJECTION]: The skill processes code snippets returned from public GitHub repositories. While these snippets could theoretically contain malicious instructions (Category 8), the skill acts as a search tool and does not execute the retrieved code or pipe it into dangerous tools. The risk is minimized by the intended use case of finding code examples.

grep-app