[GH-ISSUE #22187] Issue: Inconsistent Output Format Instructions in DEFAULT_FOLLOW_UP_GENERATION_PROMPT_TEMPLATE for LLM Follow-up Prompts

Originally created by @listenerri on GitHub (Mar 3, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/22187 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Pip Install ### Open WebUI Version v0.8.8 ### Ollama Version (if applicable) _No response_ ### Operating System Ubuntu24.04 ### Browser (if applicable) _No response_ ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior LLMs should always receive explicit output format instructions, consistently returning follow-up prompts as a JSON object with key `follow_ups`. ### Actual Behavior The prompt template includes conflicting instructions, sometimes resulting in a top-level JSON array. Backend expects an object. This causes parsing failures and missing follow-up prompts in the UI. https://github.com/open-webui/open-webui/blob/79f04379801622181ef9c591374a285eac4e1c4d/backend/open_webui/config.py#L1963-L1965 ### Steps to Reproduce 1. Inspect the default prompt template for follow-up generation. 2. Trigger follow-up suggestion and force model to return a top-level array. 3. Observe backend does not parse and UI does not display suggestions. ### Logs & Screenshots No backend error, template snippet included for reference. ### Additional Information See analysis and recommended fix above.

GiteaMirror commented

2026-04-25 09:25:16 -05:00

Owner

Originally created by @listenerri on GitHub (Mar 3, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/22187

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.

Installation Method

Pip Install

Open WebUI Version

v0.8.8

Ollama Version (if applicable)

No response

Operating System

Ubuntu24.04

Browser (if applicable)

Confirmation

I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

LLMs should always receive explicit output format instructions, consistently returning follow-up prompts as a JSON object with key follow_ups.

Actual Behavior

The prompt template includes conflicting instructions, sometimes resulting in a top-level JSON array. Backend expects an object. This causes parsing failures and missing follow-up prompts in the UI.

79f0437980/backend/open_webui/config.py (L1963-L1965)

Steps to Reproduce

Inspect the default prompt template for follow-up generation.
Trigger follow-up suggestion and force model to return a top-level array.
Observe backend does not parse and UI does not display suggestions.

Logs & Screenshots

No backend error, template snippet included for reference.

Additional Information

See analysis and recommended fix above.

GiteaMirror added the bug label 2026-04-25 09:25:16 -05:00

GiteaMirror closed this issue

2026-04-25 09:25:17 -05:00

[GH-ISSUE #22187] Issue: Inconsistent Output Format Instructions in DEFAULT_FOLLOW_UP_GENERATION_PROMPT_TEMPLATE for LLM Follow-up Prompts #35182