Hermes Atlas
Deployment & hosting

OpenDeepSeek

mouxue56-debug/opendeepseek

One-command local agent platform combining Open WebUI, Hermes Agent and DeepSeek V4

In short

OpenDeepSeek is a locally deployed agent platform that pairs Open WebUI with Hermes Agent and DeepSeek V4 by default. It can also use a custom OpenAI-compatible API, and the README is written Chinese-first.

What OpenDeepSeek does

OpenDeepSeek is a locally deployed agent platform that combines Open WebUI as the front end with Hermes Agent as the agent core and DeepSeek V4 as the default model. It also accepts a custom OpenAI-compatible API, and the README is written Chinese-first. A Hermes Smart Bridge sits between the layers, sending ordinary questions straight to a lightweight provider and routing real tasks such as files, reminders, memory and images to Hermes.

Hermes provides cross-session memory, skills, cron tasks, subagents and bridges to 16 IM platforms, and the README lists DingTalk, Feishu, WeCom, email and QQ Bot among the chat integrations. Hermes can read and write local files through a /host mount, and an optional SearXNG profile adds web search. Open WebUI supplies multi-user login, knowledge-base RAG and a PWA and desktop app. By default ports bind to 127.0.0.1; exposing it publicly requires enabling WEBUI_AUTH, setting BIND_HOST and adding a reverse proxy with HTTPS.

Key features

  • Open WebUI front end with multi-user login, knowledge-base RAG, PWA and desktop app
  • Hermes Smart Bridge that routes plain Q&A directly and real tasks to Hermes Agent
  • Hermes memory, skills, cron and subagents behind the chat
  • IM integrations including DingTalk, Feishu, WeCom, email and QQ Bot
  • Local file access through a /host mount and OCR for scanned documents
  • setup.sh commands for verify, doctor, report and fix

When to use it

  • Run a private ChatGPT-style assistant that can really edit local files
  • Queue a task at night and receive the result in a work chat app
  • Ask questions over a PDF or Word knowledge base with Chinese OCR support

Who it is for: Chinese-speaking users and teams who want a one-command, self-hosted web UI for a Hermes-powered agent.

How it fits with Hermes Agent

Hermes Agent is the agent core of the stack, handling memory, skills, scheduled tasks and subagents behind the Open WebUI front end.

How to install OpenDeepSeek

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

git clone https://github.com/mouxue56-debug/opendeepseek.git
cd opendeepseek
./setup.sh --web

Requirements: A DeepSeek API key or a custom OpenAI-compatible API, and Docker (the macOS launcher starts OrbStack or Docker for you)

Note: By default it binds to 127.0.0.1, and the README warns never to expose it publicly without authentication enabled.

FAQ

What is OpenDeepSeek?

OpenDeepSeek is a local agent platform that runs Open WebUI, a Hermes Smart Bridge and Hermes Agent with DeepSeek V4 as the default model. It also supports a custom OpenAI-compatible API.

How do I install OpenDeepSeek?

Clone the repository, change into it and run ./setup.sh --web, pasting your DeepSeek API key once. The browser then opens at http://localhost:3000, and the README lists verify, doctor and report commands for troubleshooting.

What do I need to run OpenDeepSeek?

You need a DeepSeek API key or another OpenAI-compatible API, plus Docker. For public deployment you must also enable WEBUI_AUTH, set BIND_HOST and use HTTPS behind a reverse proxy.

Similar deployment for Hermes Agent

All deployment

Related guides: How to run Hermes Agent securely · How to install Hermes Agent · Connect Hermes agents on several machines with Hermes Desktop