Appearance
Transformers
Transformers can modify, enrich, or drop events before they are persisted and printed.
Where transformers run
Transformers run inside the dispatcher pipeline for detection events:
transform → persist → (optional filter) → print
Flow events are not transformed today, and profile events are printed directly.
What a transformer can do
A transformer can:
- Update fields on the event (e.g. score, metadata, background).
- Return a new event instance.
- Drop an event by returning
nil(the dispatcher will stop processing it).
Failure behavior
If a transformer returns an error or nil, the dispatcher stops processing that event.
Built-in transformer chain: scoring
Jibril includes a scoring transformer chain (scoring) intended to evolve into an event scoring/enrichment stage.
Configuration
Enable the feature and turn on the scoring chain:
yaml
features:
- transformers
feature_options:
transformers:
scoring: trueNotes
- Transformers are disabled by default.
- Some transformers may require persisters to be enabled (e.g. to query prior events).