Search & retrieval

Bloom-filter existence index

Answers whether an id definitely does not exist in a kind without touching the stored entities.

RoadmapEngine edition 5f8ea3af

What it does

The engine maintains one growable Bloom filter per entity kind on every write and exposes an exists-in-kind check: false means the id is certainly absent, true means it may be present. Nothing in the server, command line or MCP servers calls the check, so it is an engine capability and nothing more.

How it works

Each kind has a filter created on first insert with a target false-positive rate of 1% that grows as entries are added. Puts and transactional commits insert the id after the write is applied. Deletes cannot remove an entry from a Bloom filter, so a deleted id keeps answering maybe and the false-positive rate drifts upward until the filter is rebuilt; a true answer must always be confirmed with a read.

On the disk entity profile the filter is not populated. There the check answers exactly through the disk projection instead, which makes the call cheap on a miss but turns it into a point read.

Where you reach it

  • LibraryEngine::exists_in_kind(kind, id) (Rust engine API; not exposed by the server)

Limits

  • Not reachable through any route, command or setting of the private-preview server.
  • The filter is rebuilt from the log at boot but never compacted while running, so repeated deletes and re-inserts increase false positives.
  • False positives are possible by design; false negatives are not.

Evidence

  • Tests4 integration tests cover no false negatives, bounded false positives, kind isolation and unknown kinds; a disk-profile test checks the exact answer

Test coverage is listed as present in the engine repository at this edition. This page does not report test results.

KynetraDB is in private preview

This is not available in the preview. Request a preview to discuss the current scope.