The Agent Skills Directory

[SAFE]: The skill facilitates the creation of project-level configuration files for GitHub MCP, promoting the principle of least privilege by isolating access tokens to specific projects.
[SAFE]: Uses a non-sensitive placeholder (YOUR_TOKEN_HERE) when creating the configuration file, ensuring that the AI agent does not handle or store the actual secret during the setup process.
[SAFE]: Automatically updates the project's .gitignore file to include the configuration file, which is a critical safeguard against accidental credential leakage to version control systems.
[SAFE]: Explicitly instructs the agent to avoid reading the configuration file after the user has populated it with a token, reducing the risk of accidental secret exposure in chat history.
[SAFE]: References official GitHub settings and GitHub Copilot API endpoints for configuration purposes.

github-mcp