What problem it solves

Long-horizon agents need a shared, durable world model that can be queried and updated continuously without becoming tightly coupled to the storage or execution layer. Omnigraph addresses this with typed graph data, multimodal retrieval, reviewable branches, declarative deployment configuration, and server-side policy enforcement.

Key features

• Typed ontology — Define node and edge types, keys, and constraints in `.pg` schemas. The engine enforces the declared model on every write, giving agents a shared domain structure.

  Source: README, crates/omnigraph-compiler

• Multimodal retrieval — Combine graph traversal, vector ANN, full-text search, and Reciprocal Rank Fusion in one runtime. Queries can also apply filters while assembling context.

  Source: README, crates/omnigraph-reference-engine

• Reviewable branching — Let each agent or task work on a separate branch before changes reach the main graph. The workflow includes three-way merging and time travel across the graph.

  Source: README

• Configuration declared as code — Declare graphs, schemas, stored queries, embedding providers, and policies in `cluster.yaml`. Deployment follows a plan-and-apply workflow for previewing and converging configuration.

  Source: README, crates/omnigraph-cluster

• Server-side security policies — Apply Cedar policies to every mutation at graph or server scope. Authentication, server-resolved actors, and audit tracking are handled alongside policy enforcement.

  Source: README, crates/omnigraph-policy

Quick start

Copied verbatim from the project's README.

curl -fsSL https://raw.githubusercontent.com/ModernRelay/omnigraph/main/scripts/install.sh | bash

omnigraph query  search_docs --params '{"q":"AI safety"}'
omnigraph mutate add_person  --params '{"name":"Mina"}'

omnigraph branch create --from main agent/ingest-42
omnigraph branch merge  agent/ingest-42 --into main

Who it may suit

This may suit teams building agentic memory, shared organizational knowledge, decision-trace systems, or development graphs where multiple agents and UIs need durable, governed state.

Topics to discuss

• Which typed ontology and branch-review workflows would be most useful for your agent workloads?

• How should graph, vector, and full-text retrieval priorities be balanced in a unified query?

• Which deployment and policy boundaries should be managed per graph, per branch, or server-wide?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/ModernRelay/omnigraph#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.