Skip to main content

Connectors overview

Connectors are how Premora ingests from the systems your knowledge already lives in. Every connector authenticates with its source, pulls content on a schedule or on demand, normalizes it to canonical form, and — critically — projects the source system’s permissions into Premora as entitlement snapshots.

That ACL projection is what makes Premora safe to point at sensitive systems: a result is only ever visible to someone already entitled to it upstream.

Reusable connector shapes

Premora does not build every integration as a bespoke snowflake. It supports four reusable connector shapes, and most vendor-specific connectors are profiles on top of one of them:

ShapeUsed for
document-apiPage/issue/asset content from document and work-tracking APIs.
collaboration-apiMessages, threads, and mail from collaboration systems.
structured-queryRead-only queries against databases and warehouses.
event-or-file-ingestFile drops, snapshots, and event streams.

This keeps the number of true execution engines small while supporting many systems.

What a connector preserves

  • Content — normalized to canonical markdown / JSON, deduplicated by content hash.
  • ACLs — source permissions captured as cached entitlement snapshots, enforced on every read.
  • Lineage — every materialized fact traces back to the originating source (e.g. a Confluence space, a Jira project, a Git repo).

Read-only by default

Source gateways operate with hard read-only egress — Premora reads from your systems, it does not write back to them. Structured-data connectors run against read replicas / reporting schemas and are guarded so only safe, read-only queries execute.

Where to go next