OPEN SOURCE
AgentAnywhere Core: the open foundation under the platform.
Core is the open-source agentic AI foundation we contribute to the field. AgentAnywhere is the enterprise platform on top of Core, with sovereignty, governance, support, and SLAs. The same runtime contract powers both. The boundary is explicit and documented.
What Core is, what Core is not.
AgentAnywhere Core is an open-source project. Apache 2.0 license. Public roadmap. Public issue tracker. The runtime contract that AgentAnywhere implements lives in Core; the SDKs you use to author agents are Core; the data structures that describe a flow, an agent, a model record, a tool — those are Core.
What Core does not include is the governance metadata, the audit chain, the regulatory mappings, the enterprise SSO, the sovereign-deployment tooling, and the support contracts that make the commercial product appropriate for production at a regulated enterprise. Those are intentionally not in Core. The boundary is published; nothing is rug-pulled.
If you are a developer learning agentic AI, prototyping for a side project, contributing to the field, or deploying for a use case where audit and compliance are not the constraint — Core is for you. If you are deploying at a regulated enterprise where compliance is the starting condition — the commercial AgentAnywhere is for you, and Core is the ground it stands on.
What's in Core today.
Runtime contract
The agent, flow, tool, and model interfaces that every implementation — open source or commercial — targets. Stable, versioned, semver-published.
TypeScript and Python SDKs
First-class authoring SDKs against the Core runtime contract. Same SDKs the commercial Agent Lab uses; the boundary is at the runtime, not at the SDK.
Local registry
File-based registry with the same data structure as the commercial Registry — identity, version, checksum, basic provenance. Self-host on any conformant infrastructure.
Model Hub primitives
The Model Hub data structure and the registry API for register / fetch / list / version. The full governance metadata, audit chain, and regulatory mappings live in commercial Model Hub.
Reference flows
Worked examples for common patterns — single agent, multi-agent handoff, human-in-the-loop, tool-augmented retrieval. Designed to be read, copied, and modified.
CLI and developer experience
Developer-friendly CLI for scaffolding, running, and debugging. Local development against any model provider you have credentials for.
Where the boundary sits.
The boundary between AgentAnywhere Core and commercial AgentAnywhere is intentional and documented. Code-wise, the two share the runtime contract; capability-wise, the commercial product is what you deploy when a regulator is going to look at your AI inventory.
AgentAnywhere Core (open source)
- Apache 2.0; commercial use permitted
- Runtime contract, SDKs, local Registry, Model Hub primitives, reference flows, CLI
- Self-host on any conformant infrastructure
- Community support via GitHub Issues and the Core Discord
- Public roadmap; contributions welcomed via the contribution guide
Commercial AgentAnywhere
- Sovereign deployment topology (BYO VPC / private cloud / on-premises / air-gapped)
- Full governance metadata, cryptographic audit chain, RBI FREE-AI mappings, HIPAA/GDPR documentation
- Enterprise SSO, role-based access controls, multi-tenant deployments
- TrustFabric, Guardrails, Validator, Observe — the full governance stack
- Production-grade storage backends, multi-region replication, disaster recovery
- 24×7 support, SLAs, solutions architecture, named customer success
Contribution discipline.
Core is governed as an open-source project, not as a marketing surface. Contributions are reviewed against the project's contribution guide; releases are versioned per semver; breaking changes carry deprecation cycles.
We do not accept contributions whose primary purpose is to graft commercial-edition features into Core. The boundary is intentional. The commercial product exists because regulated enterprises need compliance-grade artifacts that take a paid team to maintain — not because we are holding features hostage.
Explore Core, or talk to our team.
If you are a developer evaluating Core, the docs and quickstart live on docs.agentanywhere.ai. If you are evaluating commercial AgentAnywhere on top of Core, our team is the right path; the open-source experience is not a substitute for an enterprise pilot.