The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses a local script (install-skill-from-github.py) to download and install executable code from the author's GitHub repository (17854120992/mcp_youtube_data) and other repositories at runtime.
[COMMAND_EXECUTION]: The skill modifies sensitive configuration files for various IDEs including Cursor, Codex, OpenCode, and OpenClaw. These files are located in protected application data paths (e.g., ~/Library/Application Support/ and %APPDATA%), and modifying them allows the skill to influence IDE behavior and register external services.
[DATA_EXFILTRATION]: The skill explicitly requests an API key from the user and configures it for use with the domain mkterswingman.com. This domain is not recognized as a trusted organization or well-known service.
[EXTERNAL_DOWNLOADS]: The skill performs several external downloads to update itself and install helper skills from both trusted organizations (openai, anthropics) and the author's repository.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8) by ingesting and summarizing untrusted data from a remote server:
Ingestion points: It retrieves data from tools/list and get_patch_notes via the remote MCP endpoint at https://mkterswingman.com/mcp.
Boundary markers: There are no boundary markers or instructions to ignore embedded commands within the fetched content.
Capability inventory: The skill possesses the capability to execute shell commands and modify system configuration files.
Sanitization: No validation or sanitization of the remote tool output is performed before it is used to guide user interactions.

use-youtube-data-mcp