skaiworldwide-oss/agensgraph

AgensGraph, a high performance transactional graph database based on PostgreSQL

What it solves

AgensGraph solves the complexity of managing separate relational and graph databases. It allows developers to store both traditional tabular data and complex graph data (nodes and edges) within a single PostgreSQL-based system, enabling them to be queried together in one transaction.

How it works

It is a fork of PostgreSQL that integrates the property graph model directly into the database engine. Instead of being an extension, it treats graph data as native structures: graphs are schemas, labels are tables, and vertices or edges are rows. You can write queries using a mix of SQL, openCypher, and ISO GQL in a single statement. Because it is built on PostgreSQL, it supports standard database features like transactions, row-level security, replication, and existing extensions like pgvector.

Who it’s for

It is for developers and data engineers who need to perform complex relationship analysis alongside traditional relational data. It is particularly useful for those building AI agents, knowledge graphs, or GraphRAG applications who want to leverage the reliability of PostgreSQL while gaining the flexibility of graph traversal.

Highlights

  • Unified Querying: Mix SQL, openCypher, and ISO GQL in one query without mode switching.
  • Native Integration: Graph data and relational tables live in the same database and commit in the same transaction.
  • PostgreSQL Compatibility: Works with standard tools like pg_dump, EXPLAIN, and existing PostgreSQL extensions.
  • AI-Ready: Supports GraphRAG, vector search via pgvector, and provides integrations for frameworks like LangChain, LlamaIndex, and LightRAG.
  • Advanced Graph Features: Includes pathfinding algorithms like Dijkstra and variable-length pattern matching.

Related

  • Project
    lakekeeper/lakekeeper

    A secure, Rust-based Apache Iceberg REST Catalog that centralizes access control and table management across multiple compute engines.

  • Project
    datalevin/datalevin

    Datalevin is an open‑source, ACID‑compliant Datalog database that can be used as an embedded logical store, KV store, document store, vector search engine, and full‑text search engine. It offers a cost‑based optimizer, LMDB‑backed durability, multi‑language bindings (Clojure, Java, Python, Node.js), server mode with Raft clustering, and AI‑native features like llama.cpp embeddings. Benchmarks show it outperforms PostgreSQL, SQLite, Neo4j, and other systems on a variety of workloads.

  • Project
    karust/openserp

    OpenSERP is an open‑source SERP API and CLI that scrapes Google, Bing, Yandex, Baidu, DuckDuckGo and Ecosia, returning structured JSON/markdown results without API keys. It supports single‑engine queries, multi‑engine “megasearch” with deduplication, optional page‑content extraction, image search, proxies, caching, Docker deployment, and has SDKs for JavaScript/TypeScript, Python, an MCP server for AI agents, and an n8n node. Ideal for LLM grounding, SEO tracking, and any workflow needing live web search data.

  • Project
    emcf/thepipe

    thepipe is a Python library that extracts clean markdown, tables, and images from many file types (PDF, DOCX, PPTX, notebooks, audio/video, etc.) using vision‑language models. It returns “chunks” that can be fed to LLMs, converted to LlamaIndex documents, or used for structured extraction with Pydantic schemas. Install via `pip install thepipe-api` and add optional extras for Whisper, sentence‑transformers, GPU support, or LlamaIndex integration.