Open WebUI scheduled automations can now launch a coding agent in a saved terminal and working folder, with results sent to a chat. Scheduled agent runs no longer need a separate scheduler.
Key takeaways · AI-distilled
Open WebUI v0.13.3 lets automations run a coding AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition →: the editor picks a terminal and working folder, scheduled runs reuse them and check the terminal and agent are still available first, and agent results go to a chat, not a channel.
Terminal coding agents no longer appear in pickers such as Admin Models, Arena candidates, task-model settings or the Playground, and they cannot be saved or imported as custom models or base models.
Over plain HTTP on a local network, copy buttons now work, and clipboard prompts and Auto-Copy Response show a message when the browser blocks clipboard access instead of failing silently.
DeepSeek thinking-mode chats that use skills, knowledge or memories no longer fail with a missing reasoning error, but only when the connection's Provider is set to "llama.cpp".
The Docker image that bundles Ollama now fails to build if Ollama was not installed, instead of producing an image that cannot start Ollama.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
