Oh My Openagent

Oh My Openagent is listed under the name OmO on its own site, which describes it as an agent harness for real work rather than only code. It runs from a terminal and is installed with a one line script for macOS, Linux, WSL and Windows. The central idea is a set of trigger keywords. Adding ultrawork to a prompt tells the agent to read the codebase first, plan, prove each step and verify the result on the real surface before it stops. Typing mass ulw turns a large job into a graph of agents working in waves, and the site demonstrates a market research task that draws on thousands of sources and produces a ranking model and a presentation deck. OmO mixes models from several vendors, listing Claude, GPT, Kimi, Grok, GLM and DeepSeek families, with a tuned system prompt shipped for each model so the right profile is used automatically. Users sign in to the model providers they already use. It is aimed at developers and knowledge workers who want to hand off multi step projects such as research, experiments, backends, frontends and decks to a team of agents. The site publishes no plan prices. Model usage is paid through the user's own provider accounts.