code-yeongyu/oh-my-openagent
OmO, also cataloged as oh-my-openagent, provides an autonomous agent command-line tool built upon senpi, a fork of pi-mono. It transforms user prompts into research across thousands of sources, presentation decks, and application code. Core functionality centers on the ultrawork or ulw keyword, where the system inspects codebase context, outlines execution plans, verifies steps on target surfaces, and concludes autonomously. By invoking mass ulw, users launch coordinated multi-agent directed graphs executing parallel subtasks across specialized model configurations.
The Kibitzer module maintains memory across sessions by storing Markdown files within a Git repository. A secondary model reads historical context alongside the main agent to prevent repeating previous mistakes. Authentication via the login command integrates subscriptions for Claude, ChatGPT, Kimi, and GLM with customized system prompts. Tool execution combines multiple calls into unified runs with automatic argument repair. User configurations are managed through omo.jsonc files placed in home or project directories.
Installation scripts download and immediately execute remote code; inspect them before running. Installation uses curl -fsSL https://get.omo.dev/install.sh | bash on Unix or irm https://get.omo.dev/install.ps1 | iex on Windows PowerShell. Alternatively, package managers support bun add -g omo-ai or npm i -g omo-ai. The application launches using omo. The GitHub snapshot shows the repository is not archived and records NOASSERTION. LICENSE.md names a restricted Sustainable Use License and separate third-party licenses; read the full terms. For another coding agent, compare OpenCode; its scope differs.
Data as of : 70642 stars · 5829 forks.