Coordinate multi-harness agent sessions with Omnigent
Run Claude Code, Codex, Cursor, OpenCode, Pi, and custom agents behind one governed orchestration layer with shared sessions, policies, and sandbox options.
npx skills add agentskillexchange/skills --skill coordinate-multi-harness-agent-sessions-with-omnigent
uv tool install omnigent or pipx install omnigent, then run the Omnigent CLI to start a governed session with the harness and sandbox provider your workflow needs.Use Omnigent when an operator needs to supervise several agent harnesses on the same project without rewriting the workflow for each tool. The agent/operator can start an Omnigent session, choose Claude Code, Codex, Cursor, OpenCode, Pi, or a custom YAML-defined agent, split work across harnesses, continue the same session from another device, and apply policy or sandbox controls before risky actions run. In practice, the workflow is to launch the common server, connect the desired harnesses, assign or review agent work, and enforce approval, spend, tool, or sandbox limits while the session stays synchronized. The boundary is multi-harness agent supervision, collaboration, and governance; it is not a generic AI framework listing or a normal desktop client card.
What this skill actually does
Inputs and prerequisites: Python 3.12+, uv or pipx, optional Node.js/tmux for coding-agent harnesses, supported sandbox provider if used.
Setup notes: Install with uv tool install omnigent or pipx install omnigent , then run the Omnigent CLI to start a governed session with the harness and sandbox provider your workflow needs.
Source and verification boundary: use https://github.com/omnigent-ai/omnigent as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.