GenAI / Graph Engineer – Agentic code intelligence with Neo4j
I’m currently building an MCP server for coding agents, designed to give them a real, graph-based understanding of large codebases.
What it does:
Incremental ingestion of entire projects (minutes, not hours)
Persistent code graph (files, scopes, symbols, relationships)
Hybrid search: BM25 + semantic embeddings
Graph traversal for context expansion
(who consumes a scope, who is consumed by it, dependency depth)
Example capability exposed to agents:
const results = await brain.search({
query: "authentication middleware",
semantic: true,
boost_keywords: ["AuthService", "validateToken"],
explore_depth: 2,
summarize: true
});
With these tools, agents like Claude Code can instantly grasp a concept across a codebase, instead of painfully reading files one by one.
I’m open to:
Collaborations with teams working on Neo4j / graph / search / agentic systems
Integrating these tools into existing coding agents
Freelance / consulting opportunities around GenAI + graph-based intelligence
Portfolio & demos:
https://luciformresearch.com
Feel free to DM me if this resonates.