Elephas maintains Chronicle, the suite's long-term knowledge graph, running continuously in the background. It converts activity signals into candidates, resolves identity ambiguities, promotes high-confidence facts, and generates behavioral inferences. All writes to Chronicle go through Elephas, ensuring provenance and consistency.
Long-term knowledge graph (Chronicle) maintenance. Ingests structured signals from system journals, resolves entity identity, promotes confirmed facts, and generates inferences.
elephas.query— query Chronicle for entities, relationships, events, or inferenceselephas.ingest.journals— ingest signals from Observation, Action, and Research journalselephas.consolidate.immediate— immediate consolidation pass on pending candidateselephas.consolidate.deep— deep pass with identity reconciliation and inference generationelephas.identity.resolve— determine whether two entity records are the sameelephas.identity.merge— merge confirmed-same entities (reversible)elephas.candidates.list— list pending candidates with confidence scoreselephas.candidates.promote— promote a candidate to confirmed factelephas.candidates.reject— reject a candidate with reasonelephas.status— node/edge counts, pending candidates, last consolidation time
Ingestion logs, candidate records, merge history, and decision log. Consolidated fact records and inference outputs.
Scheduled Consolidation Passes: Immediate pass runs frequently; scheduled pass (periodic) promotes high-confidence candidates and deduplicates; deep pass (less frequent) performs full identity reconciliation and inference generation.
Read SKILL.md for operational details, schemas, and validation rules.
See references/ for detailed specifications and examples.
This project is licensed under the MIT License - see the LICENSE file for details.