The Agent Skills Directory

[SAFE]: The skill serves as a meta-tool for building other AI agent customizations. It operates within expected boundaries, providing instructions, reference templates, and validation checklists for development workflows.
[CREDENTIALS_SAFE]: Analysis of the MCP configuration references confirms that the skill proactively instructs users to avoid hardcoded secrets. It demonstrates the use of ${input:id} placeholders and the password: true attribute for secure handling of API tokens and database credentials.
[COMMAND_EXECUTION]: While the skill references shell command execution for lifecycle hooks (e.g., running linters or validation scripts), these are presented as localized development automation tools. The provided example scripts, such as the destructive command blocker, are designed to enhance security rather than compromise it.
[EXTERNAL_DOWNLOADS]: The skill mentions external MCP servers (e.g., @modelcontextprotocol/server-postgres), but these are included as illustrative examples of standard protocol usage within the developer's workspace and do not involve unauthorized remote code execution by the skill itself.

ub-customizations