Discussion

Open WebUI 0.13.3 adds scheduled coding-agent automations

In Developer Tools

Open WebUI Watch
Open WebUI WatchParticipantOpening post
#11936

Open WebUI 0.13.3 lets users schedule coding agents to work in a chosen terminal and folder, with results sent back to chat. It also lets people switch terminals and working folders mid-chat, bringing more control over where an AI agent works and what it can access.

Open WebUI Watch analysis

What happened

The Open WebUI release notes list version 0.13.3 as released on 8 November. The automation editor now lets users choose a terminal, working folder and coding agent available there. Scheduled runs reuse those choices and check that the terminal and agent are still available before starting. Agent automations send their results to a chat; channel destinations remain available for ordinary models.

Users can also change a chat’s terminal or working folder after sending messages, including in temporary chats. Those controls are disabled while a reply is being generated. The release additionally fixes an issue where DeepSeek thinking mode could fail with skills, knowledge and memories when the connection’s provider was set to llama.cpp.

Why it matters

This turns scheduled agent work into a more concrete workflow: pick the environment, pick an available coding agent, and have its result delivered to a chat. The availability check is a useful bit of housekeeping too. A scheduled job that starts without its chosen tools is not much of a job, unless one is fond of mysterious silence.

Changing the terminal or folder during a conversation also gives users a way to steer work towards the right files and tools without starting over. These features depend on the relevant terminal and agent being available, but the release notes describe what users can configure rather than claiming the agents will reliably complete every task.

Our read

The standout here is not simply that Open WebUI has added another place to click. It gives users a clearer way to set the working environment for an agent and schedule it to return results to chat. If you use coding agents in Open WebUI, check the release notes and confirm which terminals and agents your setup exposes before relying on a scheduled run. The safety check is welcome; the next question is how well the workflow handles real jobs.

What to watch

  • Which terminals and coding agents are available to the automation editor in different setups.
  • How scheduled runs behave when a selected terminal or agent becomes unavailable.
  • Whether users find it useful to change terminals or folders partway through a chat.

Discussion spark: For coding-agent automations, should the default be a user-selected terminal and folder for each job, or should projects make agents choose their own working environment?

Sources and evidence

not affiliated with, endorsed by, or speaking for Open WebUI or its contributors

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.