Write Neumann database queries across relational, graph, vector, and unified engines.
npx skills add https://github.com/Shadylukin/Neumann --skill neumann-query
Please help me install this Agent Skill. Skill: neumann-query Source: https://github.com/Shadylukin/Neumann/tree/main/.claude/skills/neumann-query Command: npx skills add https://github.com/Shadylukin/Neumann --skill neumann-query
This Skill helps developers and data engineers write syntactically correct and semantically valid Neumann database queries across relational, graph, vector, and unified engines, reducing errors, parser mis-tokenization, and wasted debugging time.
Write a Neumann query that finds engineers similar to 'user:alice' who report to 'user:bob'.
High-intent search queries and answers about installing and using this skill.
Cross-engine Neumann DB queries combine vector similarity search with graph traversal by using EMBED and SIMILAR commands for vectors alongside NODE and EDGE commands for graph relationships, connected through proper arrow and colon delimiters.
Neumann DB queries fail with colon-containing keys due to parser mis-tokenization. You must enforce quoting rules around keys containing colons to prevent the parser from interpreting the colon as a delimiter, ensuring correct tokenization and query execution.
The correct syntax for edge direction keywords in Neumann graph database queries requires specifying OUTGOING, INCOMING, or BOTH explicitly. Edge commands must use proper arrow and colon delimiters to define traversal direction and ensure the parser correctly interprets the graph relationship.
To use the SIMILAR command without syntax errors in Neumann DB, ensure proper ordering of metric and limit parameters. The SIMILAR command requires the metric specification and limit values in the correct sequence to avoid parser errors and return accurate vector similarity results.
Yes, you can cache vector embeddings and graph query results together in Neumann DB using the CACHE and VAULT commands. This supports agent memory workflows with semantic caching, allowing combined storage of EMBED results and graph traversal data for faster subsequent retrieval.
The main limitation when using reserved keywords as identifiers in Neumann database queries is parser mis-tokenization. Reserved keywords like OUTGOING, INCOMING, BOTH, NODE, and EDGE cannot be used as unquoted identifiers, requiring quoting rules to avoid syntax errors and ensure correct command parsing.
ShadylukinDiagnose Neumann errors, performance problems, and cluster faults.
ShadylukinModel and query directed property graph data with Neumann's graph engine.
ShadylukinBuild HNSW vector indexes and run similarity queries with COSINE, EUCLIDEAN, and DOT_PRODUCT metrics.
ricableRun embedded vector searches with SQL, SPARQL, and Cypher via WASM.
