Engram ExplorerInstall

Summary

Searches relationship facts for a required query within a required group, using text matching or hybrid text and semantic retrieval. Optional filters narrow facts by review status, source episode details, validity and creation dates; a relevance threshold and result limit control output, and retrieval diagnostics can return an aggregate audit record.

Inputs

• Backend — Select Embedded (Graphology, default) or Neo4j (Remote).
• Custom Storage Path — For the embedded backend, optionally set a custom directory for graph data files.
• Group ID (required) — Session or group identifier that scopes the relationship search.
• Query (required) — Text used to find matching relationships.
• Search Mode — Choose Text Only (default) or Hybrid (Text + Semantic). Hybrid combines full-text and embedding-based semantic results using reciprocal rank fusion.
• Embedding Model — Model used for semantic search in hybrid mode; the field uses the Engram Extraction LLM credential.
• Fact Review Status — Optionally restrict facts to accepted, proposed, or rejected status; Any is the default.
• Source Sender ID — Optionally restrict facts to those supported by episodes from this sender.
• Source Episode Kind — Optionally restrict facts to those supported by a selected episode kind.
• Source Trust Level — Optionally restrict facts to those supported at a selected trust level.
• Source Workflow ID — Optionally restrict facts to those supported by a specified source workflow.
• Source Execution ID — Optionally restrict facts to those supported by a specified source execution.
• Source Reference Time After — Optionally include only facts supported at or after this reference time.
• Source Reference Time Before — Optionally include only facts supported at or before this reference time.
• Limit — Maximum results to return; defaults to 10, with a declared range of 1–100.
• Min Relevance Score — Minimum relevance score from 0 to 1; defaults to 0.
• Retrieval Diagnostics — When enabled, returns an aggregate record with bounded candidate decisions and a context-budget preview; disabled by default.
• Diagnostics Candidate Limit — Maximum candidate decisions in the audit trace when diagnostics are enabled; defaults to 100, range 1–250.
• Context Preview Token Budget — Token budget for the diagnostics context preview; defaults to 1000, range 64–100000.
• Valid After — Optionally restrict facts by when they became valid, after this date.
• Valid Before — Optionally restrict facts by when they became valid, before this date.
• Created After — Optionally include results created after this date.
• Created Before — Optionally include results created before this date.

Output shape

A list of matching relationship records, or one aggregate diagnostics record when retrieval diagnostics are enabled.

The source pack provides signatures but omits method bodies, so it does not establish the exact fields in relationship records or how relevance scores are applied.

Links

• n8n community nodes installation guide