[GH-ISSUE #23781] issue: Recover full conversation history (reasoning + tools inputs/outputs) inside a tool

Originally created by @ccdv-ai on GitHub (Apr 16, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/23781 ### 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 0.8.12 ### Ollama Version (if applicable) _No response_ ### Operating System Ubuntu ### 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 I am using models capable of multi-step reasoning and tool use (Qwen 3.5 models). My workflow involves multiple tool calls sequentially within a single turn. When the LLM executes a sequence of tools (e.g., Tool 1 → Tool 2 → Tool $n$), Tool $n$ requires the data returned by Tool 1 or Tool 2 (history of the outputs). However, when inspecting the `__messages__` or `__metadata__` or `__tools__` arguments inside Tool $n$, the results/outputs of the previous tools are missing. As far as I remember, it was possible few months ago with a previous version (or maybe im missing something). The full history of tool calls/outputs should be available. ### Actual Behavior The full history of tool calls/outputs is not available. ### Steps to Reproduce 1. Create two tools: `Tool_A` (returns a string "Data from A") and `Tool_B` (prints the content of `__messages__`). 2. Use a model (Qwen 3.5) for tool chaining 3. Observe the reserved arguments of `Tool_B`. ### Logs & Screenshots None ### Additional Information _No response_

GiteaMirror commented

2026-04-25 09:46:48 -05:00

Owner

Originally created by @ccdv-ai on GitHub (Apr 16, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/23781

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

0.8.12

Ollama Version (if applicable)

No response

Operating System

Ubuntu

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

I am using models capable of multi-step reasoning and tool use (Qwen 3.5 models). My workflow involves multiple tool calls sequentially within a single turn.

When the LLM executes a sequence of tools (e.g., Tool 1 → Tool 2 → Tool n), Tool n requires the data returned by Tool 1 or Tool 2 (history of the outputs). However, when inspecting the __messages__ or __metadata__ or __tools__ arguments inside Tool n, the results/outputs of the previous tools are missing.

As far as I remember, it was possible few months ago with a previous version (or maybe im missing something).
The full history of tool calls/outputs should be available.

Actual Behavior

The full history of tool calls/outputs is not available.

Steps to Reproduce

Create two tools: Tool_A (returns a string "Data from A") and Tool_B (prints the content of __messages__).
Use a model (Qwen 3.5) for tool chaining
Observe the reserved arguments of Tool_B.

Logs & Screenshots

None

Additional Information

GiteaMirror added the bug label 2026-04-25 09:46:48 -05:00

[GH-ISSUE #23781] issue: Recover full conversation history (reasoning + tools inputs/outputs) inside a tool #35601