The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash(railway:*) tool to execute Railway CLI commands. This access is properly scoped to the necessary binary and follows the principle of least privilege for the intended deployment tasks.
[DATA_EXFILTRATION]: The railway up command transmits local source code to Railway's infrastructure for the purpose of cloud deployment. As Railway is a well-known technology service, this data transmission is considered safe and is the primary function of the skill.
[INDIRECT_PROMPT_INJECTION]: The instructions direct the agent to analyze build logs when using the --ci flag. This creates a surface where external content (build output) is processed by the agent. While this is a standard risk for deployment tools, it is noted here as the primary ingestion point for untrusted data.

railway-deploy