Back to directory

omar

A deterministic, formally specified orchestration tool for multi-agent systems, letting agents such as Claude and Codex be arranged into managed hierarchies and controlled from a web or terminal mission-control interface built on tmux.

omar

omar lets an agent such as Claude or Codex manage sub-agents in deep hierarchies, similar to a company org chart, with heterogeneous agents from different vendors collaborating as one team; a person can talk to and control any sub-agent directly, and agents can be long-running or ephemeral. Its web-based Mission Control lets a person describe a workflow in plain language, review the topology the assistant compiles it to, and confirm before anything actually runs and the diagram goes live; a legacy terminal UI is also available, built on tmux, though it does not yet implement the same deterministic model.

Other described features include integration with messaging systems such as Slack and support for computer use. Installation requires tmux, Rust, GNU Make, Node.js to build its Mission Control interface, and elan to build its companion omarc tool, plus at least one supported coding agent.

It is written in Rust, installs through a shell script, Homebrew, or from source, and documentation is offered in Chinese alongside English.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.