generated: '2026-09-12'
method: searched
source: >-
  https://developers.google.com/workspace/sites/docs/1.0/developers_guide_protocol and
  https://developers.google.com/workspace/sites/docs/1.0/reference
note: >-
  Cross-cutting runtime semantics for the Google Sites Data API, read from the API's
  own protocol guide and reference guide. This is a Google Data (GData) v1.4 API -
  Atom/AtomPub over HTTP, not JSON/REST - so several conventions that a modern REST
  reader expects (JSON envelopes, Problem Details, RateLimit headers, idempotency
  keys) simply do not exist here. Absences below are measured, not assumed.
media_type:
  request: application/atom+xml
  response: application/atom+xml
  json_supported: false
  note: >-
    All feeds and entries are Atom 1.0 documents with the GData, Sites
    (http://schemas.google.com/sites/2008), ACL, OpenSearch and batch namespaces.
    Attachments are transferred as raw media or as MIME multipart alongside the entry.
auth:
  style: oauth2-bearer
  header: Authorization
  scopes: scopes/google-sites-scopes.yml
  detail: authentication/google-sites-authentication.yml
versioning:
  style: header
  header: GData-Version
  value: '1.4'
  query_fallback: v=1.4
  detail: lifecycle/google-sites-lifecycle.yml
pagination:
  style: opensearch-offset
  params:
    - name: start-index
      description: 1-based index of the first result to return.
    - name: max-results
      description: Maximum number of entries to return in the feed.
  response_fields:
    - openSearch:startIndex
    - openSearch:totalResults
    - openSearch:itemsPerPage
  next_link: 'link rel="next" type="application/atom+xml"'
  cursor: false
  note: >-
    Offset paging. The feed carries a rel="next" link that already encodes the next
    start-index, so a client can follow links rather than compute offsets.
filtering:
  params:
    - name: q
      description: Full-text query.
    - name: updated-min
      description: Lower bound on entry update time.
    - name: updated-max
      description: Upper bound on entry update time.
    - name: published-min
      description: Lower bound on publish time (content feed only).
    - name: published-max
      description: Upper bound on publish time (content feed only).
    - name: kind
      description: >-
        Comma-separated content kinds - announcement, announcementspage, attachment,
        comment, filecabinet, listitem, listpage, webpage, webattachment, template.
    - name: parent
      description: Only entries whose parent is the given entry ID.
    - name: ancestor
      description: Only entries under the given ancestor entry ID, at any depth.
    - name: path
      description: Content under a page path, e.g. path=/path/to/page.
    - name: include-deleted
      description: Include deleted entries. Default false.
    - name: include-draft
      description: Include draft entries (owner/author only). Default false.
    - name: include-all-sites
      description: Site feed - list every site the user can view in the Workspace domain.
    - name: with-mappings
      description: Site feed - include web address mappings in each site entry.
field_expansion:
  supported: false
  note: No sparse-fieldset or expansion parameter is documented.
metadata:
  supported: true
  mechanism: >-
    Arbitrary site categories via <category scheme="http://schemas.google.com/sites/2008#tag">
    on a site entry; page-level metadata lives in Sites-namespaced elements.
request_id_tracing:
  supported: false
  note: No request-id or correlation header is documented for this API.
concurrency:
  style: etag
  mechanism: >-
    Every <feed> and <entry> carries a gd:etag attribute. Writes take an If-Match
    header (or the gd:etag attribute on the submitted entry); If-Match:* forces the
    write regardless of concurrent modification. Conditional reads use If-None-Match
    and are answered with HTTP 304 Not Modified or HTTP 412 Precondition Failed when
    the entity is unchanged.
  headers:
    - If-Match
    - If-None-Match
    - ETag
idempotency:
  coverage: none
  key_header: null
  scope: []
  retention: null
  note: >-
    The API publishes NO idempotency-key mechanism. No Idempotency-Key (or equivalent)
    header appears anywhere in the protocol guide or reference guide. Replaying a POST
    to the content feed or the site feed creates a second entry. The only replay
    protection the API offers is optimistic concurrency on UPDATE and DELETE via
    If-Match ETags, which prevents clobbering a concurrent change but does not make a
    retried CREATE safe.
  partial_mitigation:
    mechanism: etag-if-match
    applies_to:
      - PUT /feeds/content/{domain}/{site}/{entryId}
      - DELETE /feeds/content/{domain}/{site}/{entryId}
      - PUT /feeds/site/{domain}/{site}
      - PUT /feeds/acl/site/{domain}/{site}/{aclEntryId}
      - DELETE /feeds/acl/site/{domain}/{site}/{aclEntryId}
reversibility:
  grade: documented
  read_only: false
  note: >-
    Content deletion is soft - deleted entries remain retrievable through the content
    feed with include-deleted=true, and the revision feed retains historical revisions
    of content entries - so a delete is recoverable in principle. But the API
    documents NO reversal operation and NO window: there is no undelete, restore,
    untrash or rollback verb in any of the five feeds, and no retention period is
    stated anywhere in the protocol or reference guide. An agent can see that a page
    was deleted; it cannot put it back through this API.
  operations:
    - write: DELETE /feeds/content/{domain}/{site}/{entryId}
      reversal: null
      recovery_signal: >-
        GET /feeds/content/{domain}/{site}?include-deleted=true still returns the
        entry, and the revision feed retains its history.
      window: null
      window_source: null
    - write: POST /feeds/site/{domain} (create site)
      reversal: null
      window: null
      note: >-
        The site feed supports GET, POST and PUT only - no DELETE. A site created
        through the API cannot be removed through the API.
    - write: POST /feeds/acl/site/{domain}/{site} (grant sharing permission)
      reversal: DELETE /feeds/acl/site/{domain}/{site}/{aclEntryId}
      window: none-stated
      note: >-
        ACL grants are the one write with a true inverse operation in the same feed.
        No window applies - a grant can be revoked at any time.
  gaps:
    - >-
      No stated retention window for deleted content. Grading is capped at
      `documented` for exactly this reason: a reversal path exists for ACLs and a
      recovery signal exists for content, but no window is published, so an agent
      cannot know how long it has to undo anything.
dry_run:
  supported: false
  note: No preview, validate-only or dry-run mode is documented.
batch:
  supported: true
  endpoint: 'POST {feed}/batch'
  operations:
    - query
    - insert
    - update
    - delete
  mechanism: >-
    POST a batch feed to the rel="http://schemas.google.com/g/2005#batch" link on the
    content feed. Each <entry> carries a <batch:operation type="..."/>; existing
    entries are addressed by <id>, new entries by a caller-chosen <batch:id> which is
    echoed in the response so the client can correlate results.
  partial_failure: true
  partial_failure_note: >-
    "The server will perform as many of the requested changes as possible and return
    status information that you can use to evaluate the success or failure of each
    operation." A batch is not atomic.
error_envelope:
  format: gdata-xml
  rfc9457: false
  problem_json: false
  detail: errors/google-sites-problem-types.yml
  note: >-
    Errors follow the generic Google Data APIs error model, not RFC 9457. The Sites
    documentation publishes no error-code registry of its own.
rate_limit_signal:
  headers: []
  documented: false
  note: >-
    No rate limit, quota, 429 or Retry-After signal appears anywhere in the Sites Data
    API documentation. See rate-limits/google-sites-rate-limits.yml.
webhooks:
  supported: false
  note: >-
    No webhook, push-notification or change-subscription surface. The Activity feed is
    a pollable change log, not a push channel.