Skip to content
uris

The four movements

sync       resource → catalog        pull references in
analyze    content  → understanding  per item, by kind
search     catalog  → you            one index across everything
export     catalog  → resource       bytes back out

All four are reachable over /mcp and over /graphql.

sync

Section titled “sync”

Resource#sync! claims the resource, starts a Run, and enqueues SyncResourceJob.

The job iterates each_page on the resource and records a Reference per object. No bytes move. Eight of the ten types sync: six implement each_page and Caldav and Carddav inherit it from Webdav. See Resources.

app/models/resource.rb, app/jobs/sync_resource_job.rb · Sync a resource

analyze

Section titled “analyze”

Analyzer.for(item) picks the first of twelve analyzers whose handles? returns true, then runs its steps. Steps record started_at, finished_at and either result or error under reference.analysis["steps"].

app/analyzers/ · Analysis

Section titled “search”

Item.search asks SearchIndex for ids and reloads them from Postgres in index order. OpenSearch is the query path; Postgres is the system of record.

app/models/search_index.rb · Search the catalog

export

Section titled “export”

ExportItemsJob iterates the items a selector matches and copies each one’s bytes into a storage resource, recording the copy as another Reference on the same item.

app/jobs/export_items_job.rb · Export items back out

Sync and export are one form

Section titled “Sync and export are one form”

The same shape with source and destination swapped, and cataloguing as an independent flag. Four combinations follow:

source destination catalogued is
a resource — ✓ ingest — references only
a resource a storage resource — backup
a resource default storage ✓ ingest + materialize
items matching a selector a storage resource — export