by-knowledge

A persistent knowledge graph stores and queries campaign knowledge, capturing targets, scaffolds, and failure patterns to enable cross-campaign learning. It centralizes campaign data into an append-only store to quickly recall prior art and guide new designs.

Recording failure patterns and screening outcomes across campaigns requires an append-only knowledge store. You apply the graph during execution to guide decisions and at maintenance to consolidate data, ensuring future designs improve from past campaign records.

NDJSON export and import allows you to transfer campaign history by serializing structured campaign data. The knowledge graph supports NDJSON consolidation to maintain data integrity and prune duplicate records during maintenance operations.

Campaign knowledge graphs can surface prior-art at the start of a new campaign. Applying the graph at campaign start retrieves existing targets and scaffold rankings, informing initial designs with validated historical data.

This persistent memory tool enforces canonical target naming and validates campaign and failure records. This ensures structured data consistency across the append-only knowledge store during read and write operations via exposed MCP tools.

You should consolidate data and prune duplicates in a campaign knowledge graph during the maintenance phase. This prevents redundant failure patterns and campaign records from cluttering the persistent memory store, ensuring accurate cross-campaign queries.