blogger

Call the Blogger API v3 with curl + jq. The user's OAuth bearer token is in $GOOGLE_BLOGGER_TOKEN; every call needs Authorization: Bearer $GOOGLE_BLOGGER_TOKEN. Base URL: https://www.googleapis.com/blogger/v3.

Errors are {"error": {"code": ..., "message": ...}} — show them verbatim. 401 → token expired, re-connect the Blogger connector.

Always start by listing the user's blogs to get a blogId:

curl -sS -H "Authorization: Bearer $GOOGLE_BLOGGER_TOKEN" \
  "https://www.googleapis.com/blogger/v3/users/self/blogs" \
  | jq '.items[] | {id, name, url}'

Publish a post

When the user asks to publish / post / 发布 / 发出去, publish it live with ?isDraft=false (the default below) — do NOT silently save a draft and stop. Only pass ?isDraft=true when the user explicitly asks for a draft or to review before going public. After publishing, always report the returned live url back to the user.

content is HTML (not Markdown) — convert Markdown to HTML first (e.g. with pandoc -f markdown -t html or a simple converter).

• Publish a staged draft: POST /blogs/{blogId}/posts/{postId}/publish.
• Update a post: PUT /blogs/{blogId}/posts/{postId} with the same shape.

List / read posts

Gotchas

• Enable the Blogger API on the Google Cloud project backing the OAuth client, or calls 403 with accessNotConfigured.
• A 403 "Method doesn't allow unregistered callers" means the request carried no token (empty $GOOGLE_BLOGGER_TOKEN) — ask the user to (re)connect the Blogger connector, don't blame their Google Cloud setup.
• An empty {"kind":"blogger#blogList"} (no items) means the connected Google account owns no blog — tell the user to create one at blogger.com or reconnect with the account that has the blog.
• content must be HTML; passing raw Markdown will render literally.
• Paginate with &pageToken=$PAGE_TOKEN from the previous .nextPageToken.

Record the output

After you successfully publish and obtain the live result URL, call the built-in publish_artifact tool ONCE so the user can track this deliverable in My Outputs:

Use the real returned URL — never fabricate one. Call it once per published item, only after delivery is confirmed; skip it (or use status="failed") if publishing failed. See _shared/artifacts.md.