Production synchronization

Use this protected machine-client route to validate or synchronize a complete publication bundle. It is intended for an authorized source machine and a target publication owner; it is not a replacement for the single-document import workflow.

Authentication

POST /import/production requires a bearer token with all of these properties:

• Its client ID matches the source machine ID in the bundle and the source machine configured for the service.
• Its principal_type is oauth_client and its machine_owner_proxy claim is true.
• Its aud claim contains the configured production-import audience. The default is https://api.blog.service.m7.org/v1/; an installation can configure a different audience, which must match exactly.

For an existing target publication, the authenticated proxy principal also needs import.production. When the target publication does not exist, the principal needs publication.create.

Route

Route	Required fields	Optional fields	Success result
POST /import/production	bundle	apply, sync_mode	Dry-run plan or applied synchronization summary.

Request fields

Field	Required	Type	Description
bundle	Yes	Object	m7_blog_production_bundle version 1, at most 100 MiB when JSON encoded.
apply	No	Boolean	Defaults to false. Set true only after reviewing a dry run.
sync_mode	No	merge or overwrite	Defaults to merge.

The bundle must contain:

Bundle field	Required	Description
format	Yes	Exactly m7_blog_production_bundle.
version	Yes	Exactly 1.
source.machine_id	Yes	UUID that matches the authenticated client ID.
publication.source_id	Yes	Source publication UUID.
publication.slug	Yes	Target publication slug.
publication.name	Yes	Target publication name.
data	Yes	Publication data object.
integrity.data_sha256	Yes	Lowercase SHA-256 of the final compact JSON encoding of data.

data is required, but its sections are optional. Omitted record sections behave as empty arrays, and omitted settings behaves as an empty settings value. With merge, omitted sections leave target records unchanged; with overwrite, the bundle is authoritative and empty sections remove the corresponding target records.

Every record section is an array. Rows in every section except post_authors and post_tags require a unique source_id UUID. References use these source IDs and must point to a record included in the same bundle.

Data records and references

Use the following fields for a complete bundle. Fields described as optional may be omitted from a row; all other fields in the table are required when that row type is present.

Section	Required fields	Optional fields
authors	source_id, slug, name	bio, avatar_media_source_id, external_url, socials, metadata
categories	source_id, slug, name	description, summary, active, featured_media_source_id, seo, metadata
tags	source_id, slug, name	description, seo, metadata
posts	source_id, slug, title	excerpt, featured_media_source_id, category_source_id, body, body_format, status, visibility, published_at, scheduled_at, archived, archive_note
post_authors	post_source_id, author_source_id	position
post_tags	post_source_id, tag_source_id	—
pages	source_id, slug, title	body, body_format, status, visibility, published_at, archived, archive_note
redirects	source_id, source_path, destination_path	status_code
media_items	source_id, kind, state	title, alt_text, caption, metadata, deleted_at
media_renditions	source_id, media_source_id, source, state, role	asset_id, external_url, original_filename, mime_type, byte_size, width, height, duration_ms, metadata, deleted_at
content_media	source_id, media_source_id, content_type, content_source_id	usage, sort_order, alt_text, caption, metadata
domain_verifications	source_id, host, verification_method, token, status, expires_at	last_checked_at, verified_at
domains	source_id, host, verification_source_id, verification_method, verified_at	—
revisions	source_id, content_type, content_source_id, kind, snapshot	—
imports	source_id, source_name, detected_format, state	source_content, parsed_payload, parse_errors, applied_post_source_id
exports	source_id, format, state	selection, archive_filename

settings is optional. Send an object to update settings; omit it or use an empty array when it has no settings to apply. publication.featured_media_source_id, avatar_media_source_id, featured_media_source_id, category_source_id, media_source_id, content_source_id, verification_source_id, and applied_post_source_id are source-ID references; omit or set an optional reference to null when it has no target.

Posts use draft, review, scheduled, or published status; pages use draft or published; and their visibility is public, unlisted, or private. Media items use draft, ready, failed, or deleted; renditions use bigfs or external as source and reserved, ready, failed, or deleted as state. Domain verification methods are http or dns, and verification status is pending, verified, or expired. Revisions use saved or autosave; imports use markdown_front_matter or unknown and ready, invalid, applied, or cancelled; exports use markdown_zip and ready or cancelled.

Date-time fields use YYYY-MM-DD HH:MM:SS with optional microseconds, or null where allowed. The import does not preserve an exported archived_at; it applies the requested archived state and records the target-side archival time.

Dry run and apply

Start with a dry run. This is a structurally valid empty bundle; the checksum is the SHA-256 of its compact data value, {"settings":[]}. Replace the fake UUIDs with the configured machine ID and your publication source ID before sending it.

{
  "bundle": {
    "format": "m7_blog_production_bundle",
    "version": 1,
    "source": { "machine_id": "11111111-1111-1111-1111-111111111111" },
    "publication": {
      "source_id": "22222222-2222-2222-2222-222222222222",
      "slug": "engineering-notes",
      "name": "Engineering Notes"
    },
    "integrity": { "data_sha256": "845d7dd3ba293b9cb483a452d4f0c3af713d47eb9bf5e08176d6a68322d8126d" },
    "data": { "settings": [] }
  },
  "apply": false,
  "sync_mode": "merge"
}

Calculate integrity.data_sha256 only after assembling the final data value. Serialize it as compact JSON, calculate the SHA-256 digest of those bytes, and do not mutate its values or key order before sending the request.

The dry-run result has dry_run: true, applied: false, publication_id (or null when it will be created), publication_slug, publication_action (create or update), authenticated_proxy_actor_id, verified_source_machine_id, bundle_source_machine_id, sync_mode, summary, and warnings. It makes no data changes.

When the review is acceptable, resend the same bundle with apply: true. merge writes bundle records and keeps target records not represented in the bundle. overwrite makes the bundle source-authoritative and removes target publication records absent from the bundle. An applied result includes all dry-run identity fields with dry_run: false and applied: true, plus imported. imported contains the per-section imported counts, settings_saved, sync_mode, and a removed map for records deleted by an overwrite.