{
  "id": 13881988,
  "title": "Open WebUI: Creating Custom Agents",
  "url": "https://urgent.news/2026/10/12/open-webui-creating-custom-agents",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-12T05:30:00.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/admantium/open-webui-creating-custom-agents-2oa9"
  },
  "original_language": "en",
  "account": "Open WebUI is a powerful tool that can be used to create custom agents for various tasks. It evolved from a simple Ollama chat interface to a full-fledged agent interface for any OpenAI API endpoint. The agent interface can leverage custom knowledge bases, text, images, other chat references, memory, and even execute arbitrary code through a terminal feature.\n\nThe article focuses on creating a coding agent using Open WebUI version 0.9.6, which was published on June 1, 2026. The setup and configuration examples should also work with newer versions of the software.\n\nTo create a custom coding agent, you need to follow several steps. First, select a base model that is suitable for coding tasks. In this article, the model gpt-5.4-nano was chosen due to its good input and output token cost profile. The base model is configured in the Open WebUI GUI under Workspace, then Models.\n\nNext, you need to define a custom prompt for the agent. In this case, a system prompt was used that describes a senior Python full-stack engineer with a structured and rigorous work method. The prompt emphasizes the use of Python, strong typing, clear abstractions, explicit trade-offs, and a pragmatic approach to problem-solving.\n\nOnce the base model and prompt are set, you can configure the tools and skills required for the agent. In this setup, no special tools were added, and the agent was configured to remember knowledge from previous sessions. The memory feature allows the agent to persist stack choices and other relevant information during the coding session.\n\nTo execute the agent's tasks, the Terminal feature is utilized. This feature connects to a Docker container, allowing the agent to run arbitrary code. The Terminal setup involves configuring web search access, which requires creating an account with a search provider and obtaining an API key. The Terminal configuration is more involved and is discussed in a separate section of the article.\n\nAfter configuring the base model, prompt, tools, and memory settings, you can create the custom coding agent in Open WebUI. Once the agent is set up, you can initiate chats with it to generate updates to a codebase. The agent's responses will be presented in a concise, technically sharp manner, following the guidelines provided in the custom prompt. The agent will strive to provide working code when implementation is requested, state assumptions before designing complex solutions, and explain trade-offs briefly and clearly.",
  "summary": "Open WebUI evolved from an Ollama chat interface into a full-fledged agent interface for any OpenAI API endpoint. Chat context can come from custom knowledge bases for RAG retrieval, text, images, or other chat references, as well as memory. Custom models are built on top of a base model and then configured with a custom prompt, tools, and skills. Finally, the newly added Terminal feature…",
  "key_points": [
    "Open WebUI evolved from simple chat interface to full-fledged agent interface",
    "Created coding agent using Open WebUI version 0.9.6 on June 1, 2026",
    "Agent configured with Python full-stack engineer prompt and Terminal feature"
  ],
  "editors_take": "Expanding from a simple chat interface into a comprehensive agent environment allows users to build specialized, autonomous tools that integrate memory, external knowledge, and direct code execution.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}