# ModernRelay/omnigraph

Lakehouse native graph engine with git-style workflows

Repository: https://github.com/ModernRelay/omnigraph
Canonical: https://ross.abutalabs.com/products/omnigraph
Homepage: https://omnigraph.dev
Language: Rust
License: MIT
License Family: permissive
Topics: apache-arrow, datafusion, graph-database, knowledge-graph, lakehouse, lance, mcp, rust, s3, versioning, context-graph
Last push: 2026-09-02T23:18:08+00:00

## Health v2 (maintenance only)
Score: 82/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 100, release rhythm 100, longevity 10
- inputs: {"age_days": 145, "days_push": 0, "days_rel": 2, "gap_med": 5.5, "n_releases_24m": 19}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1092, forks 262 (observed 2026-09-03T02:15:06.718651+00:00)

## What it is
Omnigraph is a lakehouse-native graph database written in Rust that stores versioned graphs on object storage using the Lance format, with Git-style branching, merging, and time travel. It serves as an operational state and coordination layer for agent fleets, combining graph traversal, vector ANN, and full-text retrieval with Reciprocal Rank Fusion in a single query runtime.

## Use cases
- build a versioned knowledge graph on S3-compatible object storage
- give AI agents durable, branchable shared memory
- coordinate hundreds of agents writing to one graph without collisions
- combine graph traversal, vector search, and full-text search in one query
- manage graph schema and security policies as code
- review and merge agent data changes like pull requests
- self-host a graph database that never leaves your own storage

## When to choose
- you need Git-style branching, diffing, and merging for graph data
- you run multiple agents that must mutate a shared graph safely in parallel
- you want multimodal retrieval (vector + full-text + graph) fused in one runtime
- you require data to stay on your own S3, R2, GCS, or local storage
- you want declarative cluster, schema, and policy configuration with audit trails

## When to avoid
- you need a mature, battle-tested graph database like Neo4j for production-critical workloads
- you only need simple key-value or relational storage without graph semantics
- you want a fully managed cloud database with no self-hosting overhead
- you need broad ecosystem tooling and drivers beyond the provided CLI and HTTP API

## Facets
- artifact type: service
- maturity: active
- function: database, vector-database, search-engine, rag, agent-framework, mcp, cli, http-server, version-control, authorization
- domain: databases, large-language-models, self-hosted, developer-tools
- platform: windows, rust, self-hosted, cli
- tags: graph-database, lakehouse, git-style-versioning, knowledge-graph, lance-format, object-storage, cedar-policies, multimodal-retrieval, context-assembly, branch-and-merge, ai-agents, retrieval-augmented-generation, data-engineering, linux, macos, docker, web-server

## Member repositories
- ModernRelay/omnigraph (main) score 82

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:06.718651+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:50:20.300989+00:00, confidence not recorded.
  - readme: https://github.com/ModernRelay/omnigraph (fetched 2026-09-03T02:15:06.718651+00:00, sha eec1d413ddfd)
  - homepage: https://omnigraph.dev (fetched 2026-08-29T12:53:05.359105+00:00, sha 836962996a95)
  - site_page: https://www.omnigraph.dev/docs (fetched 2026-08-29T12:53:05.368382+00:00, sha 355f1fa9762c)
  - site_page: https://www.omnigraph.dev/docs/getting-started/install (fetched 2026-08-29T12:53:05.370223+00:00, sha 3e62b2bf2034)
  - site_page: https://www.omnigraph.dev/docs/getting-started/deployment (fetched 2026-08-29T12:53:05.371925+00:00, sha 0cd6fa0f6337)
- Data as of 2026-08-30T08:39:29.467469+00:00.