Audit relationships between new content and Knowledge Graph nodes using LLM scoring.
npx skills add https://github.com/grahama1970/fetcher --skill edge-verifier
Please help me install this Agent Skill. Skill: edge-verifier Source: https://github.com/grahama1970/fetcher/tree/main/.agents/skills/edge-verifier Command: npx skills add https://github.com/grahama1970/fetcher --skill edge-verifier
Verifies relationships between new memory artifacts and an existing Knowledge Graph by running recall and audit steps, enabling automatic edge creation with rationales and weights.
Verify and link the edges for a given source text by running the edge-verifier tool on your memory artifact.
High-intent search queries and answers about installing and using this skill.
Recalling related lessons using KNN search and auditing relationships with an LLM verifies new content against an existing Knowledge Graph. This scores candidates and creates edges of type verifies, contradicts, or related.
LLM auditing evaluates recalled candidate relationships to produce stance labels like verifies, contradicts, or related. It generates a rationale and weight for each edge before linking it into the graph.
Yes, edge verification works with ArangoDB by using upsert-based graph updating to link verified edges. It attaches provenance metadata for auditability and future reasoning within your graph memory.
Yes, you can schedule automated verification workflows for continuous graph enrichment. This automatically processes episodic memory sources and lesson nodes to link new artifacts over time.
The best way to stratify KNN search results is to score candidate edges during the recall phase and group them before LLM auditing. This ensures deterministic search and accurate relationship verification.
Yes, you need graph_memory and python-dotenv dependencies to run edge verification. The workflow also requires the requests and scillm libraries to execute KNN recall and LLM auditing successfully.
michaeldowlingStore and retrieve typed memory nodes as a queryable graph in Neo4j.
matsenExtract citation keys from .tex files and create knowledge-graph edges.
RigohlCoordinate hybrid semantic and full-text search across multiple engines via FFI.
Sonder4Flash, read, and write STM32 memory via ST-Link CLI tools.
