skills.rest
liliang-cn
liliang-cnCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

cortexdb

Provide embedded vector search, graph knowledge, and memory in a single SQLite file.

247|4|Updated Aug 7, 2025
Search Tags:#mcp#memory#embedding#vector-search#cortexdb#graph-knowledge
One-click install
npx skills add https://github.com/liliang-cn/cortexdb --skill cortexdb
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: cortexdb
Source: https://github.com/liliang-cn/cortexdb/tree/main
Command: npx skills add https://github.com/liliang-cn/cortexdb --skill cortexdb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

CortexDB provides an embedded, pure-Go storage kernel that unifies vector search, graph knowledge, and memory for AI agents, all within a single SQLite file.

Core Features & Use Cases

  • Vector search with multiple index types (HNSW, IVF, Flat) and similarity metrics (cosine, dot, euclidean)
  • Hybrid search that fuses vector similarity with full-text search (FTS5) for robust recall
  • Built-in memory system (Hindsight) and knowledge graph capabilities for multi-hop reasoning
  • MCP Tool Calling support and no-embedder mode options for LLM-driven workflows
  • Zero external dependencies and easy cross-compilation as a pure Go solution on SQLite

Quick Start

Install CortexDB in a Go project and start storing embeddings, querying with semantic search, and building graph-backed knowledge contexts with a single SQLite file.

Frequently Asked Questions about cortexdb

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add vector search and knowledge graph memory to my Go AI agent?▼

You can embed vector search, graph knowledge, and memory into Go AI agents using a pure-Go storage kernel that unifies these capabilities within a single SQLite file.

Does pure Go vector search support hybrid full-text and semantic retrieval?▼

Yes, hybrid search fuses vector similarity with SQLite FTS5 full-text search, providing robust recall for queries that need both semantic and keyword matching.

Can I run vector search in Go without an external embedding service?▼

Yes, a no-embedder mode allows storing and querying with externally provided vectors, enabling semantic search workflows without integrating a dedicated embedding model.

What index types are available for vector search in a Go application?▼

The vector search implementation supports HNSW, IVF, and Flat index types, offering multiple similarity metrics including cosine, dot product, and euclidean distance.

How does MCP Tool Calling integrate with vector and graph knowledge storage?▼

MCP Tool Calling support enables LLM-driven workflows by allowing models to interact with the vector search, memory, and graph knowledge context construction features directly.

What are the limitations of using a single SQLite file for AI agent memory?▼

Relying on a single SQLite file with HNSW, IVF, and FTS5 indices means storage scales with the local filesystem, requiring a Go toolchain and modernc.org/sqlite dependency for cross-compilation.

Related Skills

Explore More Software Engineering Skills→
zepdbzepdb
43

zeppelin

Index and search text with Zeppelin vector store and BM25 full-text indexing.

Official
Advanced
ruskibeatsruskibeats

rag-evidence-contract-testing

Validate retrieved graph edge evidence structure for LLM RAG context assembly.

Community
Intermediate
neo4j-contribneo4j-contrib
101

neo4j-graphrag-skill

Build GraphRAG retrieval pipelines on Neo4j with neo4j-graphrag retrievers.

Official
Advanced
ionmidoriionmidori

performing-similarity-search

Implement similarity search with Pinecone and Google Generative AI embeddings.

Community
Advanced