The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs shell commands by interpolating user-provided variables such as agent names, usernames, and project IDs. Evidence: Commands like gh repo create <username>/<agent-name> and railway add -s <agent-name> in SKILL.md directly insert user strings into the shell. Ingestion points: user-provided placeholders (, , ); Boundary markers: none; Capability inventory: subprocess calls to gh, railway, git, and curl; Sanitization: none. This creates a surface for command injection if the inputs are not properly sanitized by the agent.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of external command-line tools from well-known services. Evidence: The instructions include npm install -g @railway/cli and brew install gh. These are legitimate tools provided by Railway and GitHub respectively and are documented as standard prerequisites for the deployment task.

railway-deploy