llms.txt
Omnigraph
Open source graph database for AI agents and context graphs. Object-storage native, with Git-style branching and a typed ontology.
Omnigraph keeps a typed property graph as Lance tables in storage you own: S3-compatible object storage, Azure Blob Storage or a local directory. Agents and apps read and write it through the omnigraph CLI, an HTTP server or MCP. Writes land on branches that merge like Git, and every commit stays queryable. Omnigraph is MIT-licensed.
To set Omnigraph up for a user, follow the onboarding guide below. The CLI also carries an agent skill that matches its version: run omnigraph skill show. Every docs page is available as markdown: add .md to its URL, or send Accept: text/markdown.
Start here
- Onboarding guide: For coding agents: install the CLI, build a first local graph, query it, then pick the next step for what the user wants to build.
- Install: The installer downloads a release archive, verifies its SHA-256 checksum, and installs the binaries in ~/.local/bin.
- Quickstart: This walks the core loop end to end: define a schema, initialize a graph, load data, query it, and use a branch. It uses a local file-backed graph.
- MCP: omnigraph-server serves the Model Context Protocol at /mcp, so Claude, ChatGPT, Cursor and other MCP clients can work with a graph with a URL and the caller's…
- All docs in one file: Every docs page below, as markdown, in one file.
Docs
- Overview: Lakehouse native graph engine with git-style workflows
- Getting started
- Quickstart: This walks the core loop end to end: define a schema, initialize a graph, load data, query it, and use a branch. It uses a local file-backed graph.
- Install: The installer downloads a release archive, verifies its SHA-256 checksum, and installs the binaries in ~/.local/bin.
- Deployment: OmniGraph can keep a cluster on a local filesystem, S3-compatible object storage, or Azure Blob Storage.
- Concepts
- Schema
- Schema Language (
.pg): A schema declares the node and edge types a graph accepts. OmniGraph validates the same schema and constraints for mutation queries, loads, and branch merges. - Blobs: Use Blob properties for bytes that should not be projected through .gq. Blob values can contain graph-managed bytes, an external URI reference, or null.
- Schema Language (
- Queries & search
- Query Language (
.gq): A .gq file contains named, typed queries. - Session settings: A file may open with settings lines, before its declarations or its one statement.
- Explain: explain before a read declaration returns the query's v2 plan instead of its rows, regardless of the selected execution engine; before a load csv statement it…
- Explaining a load statement: The statement accepts every load option. It reads metadata, but no payload, and publishes nothing.
- Engine catalogue: The engine describes itself in GQ.
- Traversal and match patterns: Omitted bounds always mean exactly {1,1}. Explicit {1,1} is equivalent.
- Diagnostics: Every refusal of a query carries four things: a stable code, where the failure is, what was expected or violated, and one concrete fix.
- Mutations and Loads: Mutation statements live inside a named .gq query. Run them with omnigraph mutate.
- Conditional mutations: The condition compares the effective head of the target branch.
- Search: OmniGraph combines vector, full-text, and graph patterns in one .gq query.
- Embeddings: OmniGraph can turn query text into a vector for nearest($document.embedding, $text).
- Query Language (
- Branching & history
- Branches, Commits, and History: A branch is an isolated, durable graph history. Use branches to prepare and review a multi-step change without exposing intermediate results on main.
- Merging Branches: A merge integrates one branch into another and publishes the target change atomically across the whole graph.
- Changes, Diffs and Change Feeds: OmniGraph can describe what one commit changed, compare any two points in history, review what a branch changed, or deliver an ordered feed of changes on a…
- Operations
- Maintenance: Follow the command-specific concurrency requirements below.
- Recorded operations: A cluster server can run optimize on a graph's main itself, so a scheduler or an operator starts it with one HTTP request and looks the result up later by an…
- Upgrade from 0.13 to 0.14: Finish pending deployments and resolve uncertain writes with the old CLI. Stop application traffic, servers and writers.
- Upgrading OmniGraph: Normal open accepts storage formats v14 to v17 and never migrates a graph: a v14 or v15 branch is stamped v16 at its first publish, and only the storage…
- Authorization and actors: OmniGraph uses Cedar policy bundles to authorize graph, server, and cluster configuration actions.
- HTTP server: omnigraph-server serves every healthy graph in one applied cluster under /graphs/{graph_id}/…. It has no single-graph boot mode.
- Server logs and traces: The server writes text logs to stdout by default. Set OMNIGRAPH_LOG_FORMAT=json for structured logs.
- MCP: omnigraph-server serves the Model Context Protocol at /mcp, so Claude, ChatGPT, Cursor and other MCP clients can work with a graph with a URL and the caller's…
- Troubleshooting: CLI failures include a human-readable message. Add --json where supported to receive structured fields suitable for automation.
- Storage failures and restart: A served write sends many storage requests: data files, a transaction file, and last the conditional create of the graph's next manifest version, which…
- S3 cluster roots: A cluster whose server is stopped has no served route for deployment, recovery or maintenance.
- Clusters
- Operating a cluster: An OmniGraph cluster is a declarative bundle of graphs, schemas, stored queries and authorization policies.
- Cluster configuration reference: --config defaults to the current directory. Unknown fields and duplicate YAML keys are errors so that misspelled intent is never ignored.
- CLI
- CLI guide: Upgrading an existing installation? Start with the 0.13 → 0.14 checklist for before/after YAML and command examples.
- CLI reference: Use omnigraph --help or omnigraph
--help for exact installed flags. - Operator configuration: Use profiles for service/organization access and targets for resource access. This is target and credential protocol 2, with YAML version: 1.
- Open Studio: The command reads the operator configuration and nothing else: it sends no request to the server, reads no credential and starts no process that serves a graph.
- Managed data access: This guide covers data credentials for an existing managed cluster.
- Discover, configure and retire a managed cluster: These commands use your managed service identity and permissions. They do not open graph storage or require cloud credentials.
- Onboard and remove organization members: Invite teammates and manage organization membership through the managed CLI.
Product
- Omnigraph: Open source graph database for AI agents and context graphs. Object-storage native, with Git-style branching and a typed ontology.
- Omnigraph Studio: Omnigraph Studio, the Knowledge IDE: a GPU-rendered desktop app to visualize your ontology and graph, inspect knowledge, and review changes before they merge. For macOS, Windows and Linux.
- Omnigraph Cloud: Omnigraph, managed for you, with an enterprise-grade control plane: SSO, isolation, policy, audit and declarative config. Join the private-preview waitlist.
- Source on GitHub