Awesome AI Agents › Multi-Agent Frameworks

qntx/ovo

⭐ 568 Rust repository created 2026-01-28

Ovo is a Rust library that provides a multi-agent runtime kernel intended to be embedded inside a host application rather than run as a standalone service. It is published on crates.io and was formerly released under the name machi, with the GitHub repository renamed accordingly. The kernel is described as dual-mode. In the dynamic mode agents are spawned at run time and interact through the kernel, which suits open-ended behaviour where the set of participants is not known in advance. In the second mode behaviour is written as a workflow in Rhai, an embedded scripting language for Rust, and executed against a journal, so each step is recorded and a run can be reconstructed or resumed from its log instead of relying on in-memory state. Repository topics point at the surrounding ecosystem the authors target: the agent-to-agent and Model Context Protocol interoperability standards, the x402 payment scheme for machine-to-machine transactions, and zero-knowledge proof machinery including zk-SNARKs and zkVM execution, which suggests an interest in agent actions that can be verified or paid for without trusting the executor. Licensing is the usual Rust dual MIT and Apache 2.0 arrangement. The published README is short and consists mainly of badges, licence terms and project branding from QuantX, the organisation behind it, so the concrete API surface, the journal format and the degree of protocol support have to be read from the crate documentation and source rather than from the repository front page. Developers who want agent orchestration compiled into a Rust binary, with deterministic replay and no Python runtime in the deployment, are the audience; anyone evaluating it should expect to read the code.

https://github.com/qntx/ovo

a2aagentautonomous-agentsmcpmulti-agent-runtimerustworkflowx402zkvm