Omnigent

Omnigent coordinates AI agents for developers and teams who want to combine coding assistants, control their access, and supervise work in shared sessions. It runs locally or on a self-hosted server, with remote cloud sandboxes as another option.

• Agent coordination: Combine Claude Code, Codex, Cursor, OpenCode, Hermes, Pi and custom agents, delegate tasks, or have one agent review another's work.
• Action policies: Enforce spending caps, model routing and tool access rules, with approval requests for risky actions.
• Sandboxed execution: Run agents in local or remote sandboxes; on Linux and macOS, restrict filesystem and network access and provide credentials without exposing them to the agent.
• Shared sessions: Invite teammates to review, comment and steer live work, with conversation history, sub-agents, terminals and files kept in sync across terminal, browser and mobile access.
• Custom agents: Define agents with prompts, Python tools, MCP servers and sub-agents, or use the built-in Polly coding orchestrator and Debby model debate.

Omnigent is open source under Apache 2.0 and has a native macOS app. It supports provider API keys, Claude and ChatGPT subscriptions, and compatible gateways such as Ollama, vLLM and LiteLLM. Local model connections are available through those gateways; cloud model providers and remote sandboxes process work outside your machine. Omnigent collects anonymized usage telemetry by default, with an opt-out.