Back to directory

OmniBot

An on-device Android AI agent, built natively in Kotlin and Flutter, that operates terminals, manages system tasks such as alarms and calendar events, and can bridge to a Codex CLI session running on a desktop machine.

Visit website (opens in a new tab)omnibot.omnimind.com.cn
OmniBot

OmniBot is an on-device AI agent for Android that goes beyond chat with an understand, decide, execute and reflect loop. It combines an extensible tool ecosystem, including installable Skills, a bundled Alpine Linux environment, browser access, MCP connections and Android system-level tools, with system actions such as scheduled tasks, alarms, and creating, querying or updating calendar events, plus a short- and long-term memory system backed by embeddings.

Skills install by sending OmniBot a repository link, and a recommended skill collection is maintained separately; scheduled tasks run as full subagent flows capable of completing an assigned task, while alarms remain simple reminders. A remote Codex bridge lets OmniBot connect to a Codex CLI session already logged in on a separate PC or Mac, paired over the local network by scanning a QR code from the phone.

It is built with native Android Kotlin and Flutter, and the project also has a separate iOS and macOS counterpart called ViaVera.

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.