const README_MD: &str = "# {{name}}\n\nA [Materialize](https://materialize.com) project managed by mz-deploy.\n\n## Project structure\n\n- `models/` \u{2014} SQL model definitions organized by database and schema\n- `clusters/` \u{2014} Cluster definitions\n- `roles/` \u{2014} Role definitions\n- `project.toml` \u{2014} Project configuration\n\n## Profiles\n\nA *profile* is a named environment (e.g. `dev`, `staging`, `prod`). Each\nprofile has two halves:\n\n- **Connection details** \u{2014} host, port, credentials, TLS settings \u{2014} live in\n `~/.mz/profiles.toml` and are shared across all your projects.\n- **Project-side overrides** \u{2014} database/cluster suffixes, psql-style\n variables, AWS secret provider \u{2014} live under `[<profile>]` in\n `project.toml`.\n\nMost commands resolve the active profile in this order: `--profile <name>`\nflag, `MZ_DEPLOY_PROFILE` env var, then your per-checkout default. Commands\nthat don\'t connect to a database \u{2014} `compile`, `test`, and `explain` \u{2014} work\nwithout a profile selected as long as your SQL doesn\'t reference any\n`:variables`. If it does, you\'ll be prompted to set one.\n\nSet your per-checkout default once:\n\n```sh\nmz-deploy profile set dev\n```\n\nThis writes `.mzprofile` (gitignored, per-developer) so each teammate can\nchoose their own default without touching shared config. Inspect with:\n\n```sh\nmz-deploy profile list # all available profiles, with the default marked\nmz-deploy profile current # which profile will be used and where it came from\n```\n\n## Agent skills\n\nAgent skills from [MaterializeInc/agent-skills](https://github.com/MaterializeInc/agent-skills) help AI agents work with Materialize.\n\nInstall them with:\n\n```sh\nnpx -y skills add MaterializeInc/agent-skills -a universal -a claude-code --project\n```\n\nLater, update to the latest version with:\n\n```sh\nnpx -y skills update\n```\n";