The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Refers to official Railway installation scripts located at railway.com and agents.railway.com. These are trusted vendor endpoints.
[COMMAND_EXECUTION]: The skill uses the subprocess module to wrap the railway CLI for various management tasks and uses SSH to perform performance introspection on remote database instances.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves and analyzes raw database logs and performance metrics from Railway's API. This data is external and untrusted, but the skill is designed to use this information for diagnostics under agent supervision.
[CREDENTIALS_UNSAFE]: The skill reads the authentication token from the local Railway config (~/.railway/config.json). This is the standard method for CLI tool integrations, and the scripts follow best practices to avoid leaking this token in system process lists.

use-railway