Selection and privacy planning

Notrios can dry-run the exact note/resource boundary for a future backup, subset transfer, or publication handoff without writing files or changing the database.

curl -s -X POST http://127.0.0.1:8080/api/v1/selection/plan \
  -H 'Content-Type: application/json' \
  -d '{
    "target":"publication_handoff",
    "selection":{
      "notebook_ids":["nb_public"],
      "include_notebook_descendants":true,
      "tags":["publish"],
      "match":"all"
    },
    "policy":{"max_resource_bytes":25000000},
    "detail_limit":100
  }' | jq

The three targets have different defaults:

Target	Default behavior
`full_archive`	All current and trashed notes; preserve provenance, private metadata, source bundles, and links.
`subset_transfer`	Requires selectors; current notes; preserve provenance/source bundles; report cross-boundary links.
`publication_handoff`	Requires selectors; exclude `private`, `draft`, `confidential`; strip provenance/private metadata/source bundles; plan private/broken links as plain text.

Selectors can use recursive notebook IDs, tags, the normal query language , and up to 1,000 explicit note IDs. Values in one selector type are ORed. Set match to all when the populated selector types must all match.

The response contains content-free IDs, revision IDs, exact resource hashes, link classifications, source-bundle hashes allowed by policy, exclusions, metadata decisions, warnings, complete counts, and manifest_sha256. It never contains note bodies, resource bytes, source metadata JSON, source URLs, raw source-bundle keys, or local storage paths. Bundle source/item keys are hashed. Detail arrays are capped; when truncated is true, the counts and digest still describe the complete bounded plan.

Planning is read-only. It does not create archive v2, publish a site, rewrite links, copy resources, or invoke movenotes-v3; those are later explicit steps after reviewing this report.