The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous commands for interacting with the system shell to manage the Caddy process and system services.
Commands include ~/.local/bin/caddy reload for configuration updates.
Uses launchctl (macOS) and systemctl (Linux) for service lifecycle management.
[EXTERNAL_DOWNLOADS]: The documentation references external sources for building and extending the tool.
Instructions include using xcaddy to build Caddy with the caddy-dns/vercel plugin from GitHub.
Mentions a companion skill dev-serve located at an external hub (clawhub.com).
[PRIVILEGE_ESCALATION]: The skill requires elevated privileges to perform its core functions.
Uses sudo launchctl on macOS to manage the Caddy LaunchDaemon, which is required to bind to privileged port 443.
Employs sudo setcap on Linux to grant the Caddy binary permission to bind to low-numbered network ports without running as root.
[INDIRECT_PROMPT_INJECTION]: The skill instructions involve interpolating user-supplied data (application names, subdomains, and port numbers) into configuration files and shell commands.
Ingestion points: Placeholders like appname, YOUR_DOMAIN, and 31XX in SKILL.md and reference.md.
Boundary markers: None explicitly defined for the interpolation process.
Capability inventory: The skill can write to configuration files (Caddyfile, .env, plists) and execute service management commands via shell.
Sanitization: No specific sanitization or validation steps are provided for the user-supplied placeholders.
[CREDENTIALS_UNSAFE]: The skill provides guidance on managing sensitive authentication material for the Vercel API.
Recommends storing Vercel API tokens in environment files (e.g., ~/.secrets/vercel/.env) or service configuration files (plist env vars).
Includes instructions for approving device pairing requests via openclaw devices approve.

caddy