Update to Aveiro

{
  "description": "Write, edit and publish pages on an Aveiro website, such as blog posts, docs pages and landing pages. Use when the user asks to draft a page, change an existing page, add images to a page, or publish site changes.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 384
    }
  ],
  "name": "write-and-publish-pages",
  "skill_md_contents": "---\nname: write-and-publish-pages\ndescription: Write, edit and publish pages on an Aveiro website, such as blog posts, docs pages and landing pages. Use when the user asks to draft a page, change an existing page, add images to a page, or publish site changes.\n---\n\n# Write and publish pages\n\nPages are drafts until they are published. Explicit instructions from the user take priority over these steps.\n\n## Before writing\n\n1. Find the site with `aveiro_list_sites`, unless its ID is already known in this conversation.\n2. Call `aveiro_authoring_guide` once per conversation, before the first `aveiro_create_page` or `aveiro_update_page`, and follow it. Pages use Aveiro's MDX dialect: content pages are headings and prose with registered components only where they help; landing pages are built from page sections. Never write raw HTML layout.\n3. Ask for what you can't know: the topic, the audience, key facts, links. Don't invent product details, numbers, quotes or dates.\n\n## Change an existing page\n\n1. Find it with `aveiro_list_pages`, then read it with `aveiro_get_page`.\n2. Call `aveiro_update_page` with the complete new body in `content`; it replaces the whole body, so keep every part the user didn't ask to change. Pass the revision from `aveiro_get_page` as `if_match`, so an edit made in the meantime isn't overwritten.\n3. To change only the title, summary, tag, image or display date, send `metadata` and leave `content` out. `publishedAt` must be a real date (YYYY-MM-DD).\n\n## Add a page\n\n1. Pick a path inside the right folder, for example `/blog/spring-update.mdx`. If the folder doesn't exist, create it with `aveiro_create_collection` first.\n2. Call `aveiro_create_page` with the path, the MDX body and metadata (at least a title and a summary).\n\n## Images\n\n- Check `aveiro_list_media` first and reuse an existing image's URL.\n- For a new image from a public HTTPS URL, call `aveiro_import_media_from_url` and embed the `mdx` it returns.\n- For a screenshot of a real web page, call `aveiro_media_create_screenshot` with `estimate_only` true first, tell the user the credit cost, and render only after they agree. Then poll `aveiro_media_get_screenshot` until it is done.\n- Never invent an image URL.\n\n## Publish\n\n1. Show what is waiting with `aveiro_list_changes` and summarize the changes.\n2. Call `aveiro_publish_site` only when the user has asked to publish in this conversation. Name the pages that will go live, and pass their `content_ids` so nothing else is published by accident.\n3. If the site has translations, ask whether to publish them too and set `locale_scope` to match.\n\nDelete a page (`aveiro_delete_page`) only when the user asks for it and confirms which page.\n"
}