Show HN: TabyAgent – A lighter and easier alternative to OpenClaw/Hermes(github.com)
github.com
Show HN: TabyAgent – A lighter and easier alternative to OpenClaw/Hermes
https://github.com/gpdir16/tabyAgent
0 comments
—
https://github.com/gpdir16/tabyAgent
—
It runs exclusively within Docker without requiring host mounts and allows you to chat via Telegram.
I’ve used both OpenClaw and Hermes in the past, but I decided to build this myself because of their instability and, in the case of OpenClaw, its heavy footprint and numerous bugs.
It was intentionally designed to be simple, so it lacks the complex features found in OC or HA—such as support for various messengers, image generation, and voice calls—but I consider this a strength because it makes it easier to maintain and helps keep it lightweight.
As of this writing, version 0.3.1 has been released and is available on GitHub under the AGPL 3.0 license: https://github.com/gpdir16/tabyAgent
While all the features I personally need are already implemented, please let me know if you have any requests for additional features or feedback while using it.
# Key Differences
- Uses approximately 9 times less RAM than OpenClaw (OC ~3GB, HA ~600MB) - Runs in Docker without host mounting, reducing the risk of host corruption (OC requires host connection, HA requires host connection) - Interactive and easy setup via a single installation script and the Telegram bot (OC/HA setup is complex) - Proper table rendering within Telegram (not supported by OC/HA) - More cost-effective due to relatively lower token usage and call frequency (minimizes background and non-essential calls)
# Supported Features
- Browser Control: Built-in support for browser-use and Playwright - Recurring Tasks (Cron Jobs): Agents autonomously configure and automatically execute tasks at specific times - Built-in Skills: Importing school meal plans/schedules, searching for skills, creating skills, etc. - File and Terminal Tools: Capable of performing nearly any task - Various Providers: OpenAI, OpenRouter, Synthetic, etc. - Automatic Memory: Intelligently adds, deletes, or updates context during conversations or when starting new sessions - Markdown: Supports Telegram API formatting; tables are rendered as images for transmission - Real-time Status: Model and tool call counts, context window, output status, active tools, elapsed time, etc. - Infinite Context: Retains recent messages while compressing only old or unimportant ones - X Posts: Retrieve Twitter posts without an X API key (using FxEmbed) - And much more
# Supported Platforms
Supported on any platform with Docker.
For macOS and Linux, you can install it with a single line using the automatic installation script. (Windows requires manual installation)