graphify · git:20261028.8d1d6ce · 2026-10-28 · sha256 13fad1f04c7b3503
graphify git:20261028.8d1d6ceA
Immutable. This exact content never changes and is served at /api/v1/blob/13fad1f04c7b3503.
--- name: graphify description: "Knowledge graph hybride Neo4j + Qdrant. Stocker et retrouver concepts, decisions et entites ; traversee de graphe et recherche semantique ; indexer la sortie de recherche ; seed depuis le vault Obsidian. Declencheurs : stocke dans le graphe, qu'est-ce qui est relie a X, retrouve les concepts lies, graphify_store, graphify_retrieve." --- # SKILL -- graphify ## Trigger conditions Use this skill when the conversation involves: - Storing knowledge, concepts, decisions, or entities in the graph - Retrieving related concepts via graph traversal or semantic search - Querying "what is connected to X", "what decisions relate to Y" - Indexing research output from deep-research-core - Seeding the graph from Obsidian vault notes - Running a LangGraph research -> extract -> store -> reflect loop ## What graphify does graphify is a hybrid knowledge graph combining Neo4j (structural) and Qdrant (semantic). Write path: 1. Receive content (text, metadata, node type) 2. Create/update Neo4j node with relationships 3. Embed content -> Qdrant vector (same ID as Neo4j node) Read path: 1. Query arrives (natural language or structured) 2. Parallel: Neo4j graph traversal + Qdrant vector search 3. Merge + rank results 4. Return enriched context to caller ## Node types Concept | Entity | Task | Skill | Agent | Source | Session | Decision ## MCP tools (when graph-server is active) - graphify_store(node_type, title, content, relationships[]) - graphify_retrieve(query, mode="hybrid", limit=10) - graphify_relate(from_id, rel_type, to_id) - graphify_ping() ## LangGraph state mapping Concept -> state["knowledge"] Task -> state["tasks"] Decision -> state["context"] Source -> state["sources"] ## Dependencies - Neo4j port 7687 : docker compose up -d neo4j - Qdrant port 6333 : docker compose up -d qdrant - NEO4J_URI, NEO4J_USER, NEO4J_PASSWORD, QDRANT_URL in .env ## Status v0.1 scaffold -- planning only. Implementation in Phase 3. Use validate_repo.py --show-stubs to track progress.