What problem does it solve? Launched sites silently break: TLS certificates expire, hosts drop security headers after config changes, and bad deploys blank the homepage — all invisible until a visitor or attacker notices. This Skill turns those checks into a mechanical daily routine with automatic GitHub issue alerts. ## Core Features & Use Cases - Daily health checks: Verifies every configured route returns HTTP 200, the homepage contains an expected marker text, required security headers are present, and the TLS certificate has enough days before expiry. - Config-driven automation: Drop a watchdog.json into a project directory and a GitHub Actions cron picks it up daily at 07:00 UTC, opening or commenting on a GitHub issue when checks fail. - Three-verdict honesty: Distinguishes pass, site failure, and NOT CHECKED (when the runner itself has no network route), using a control host and IPv4/IPv6 retries so a blind monitor never reports a false outage. - Use Case: After launching a marketing site to production, add a watchdog.json with its URL, routes, and marker text; the next morning any cert, header, or uptime regression appears as a GitHub issue. ## Quick Start Ask the agent to run the site-watchdog check locally against your project's watchdog.json to verify the site's routes, headers, and certificate.
