The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard shell commands and Node.js one-liners to read/write local configuration files (.mcp.json) and verify API connectivity. These operations are restricted to the local project directory and standard user paths.
[EXTERNAL_DOWNLOADS]: The skill interacts with api.grep.ai, which is the official API endpoint for the 'parcha-ai' vendor resources. These network operations are used for legitimate service verification and follow the expected behavior for a tool-integration skill.
[CREDENTIALS_SAFE]: The skill manages authentication tokens for the Grep service. It correctly advises users to use long-lived API keys instead of short-lived session tokens and explicitly warns against committing these credentials to version control. It stores credentials in the standard .mcp.json file used by the agent platform.
[DATA_EXPOSURE]: The skill reads from ~/.grep/session.json to check for existing sessions. This is a local read of the vendor's own application data to facilitate a smoother setup process and does not involve exfiltrating data to unauthorized third parties.

grep-mcp