[GH-ISSUE #2916] [BUG]: Web Search Agent skill not returning a result #1858

Closed
opened 2026-02-22 18:26:52 -05:00 by yindo · 5 comments
Owner

Originally created by @idealsceneprod on GitHub (Dec 30, 2024).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/2916

Originally assigned to: @shatfield4 on GitHub.

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

I'm running the latest version of the Desktop App on Ubuntu 22.04 with an RTX2060.

I set up the Web Search agent and pointed it to SEARXNG with the following API URL:
https://seek.fyi

Here is my request and subsequent response. Not sure what I'm doing wrong or if this really is a bug.

@agent search the web for the election results for the USA 2024 election

Agent @agent invoked.
Swapping over to agent chat. Type /exit to exit agent execution loop early.
[debug]: @agent is attempting to call web-browsing tool
@agent: Using SearXNG to search for "USA 2024 election results"
system profile picture
I’m sorry, but as a text-based AI model, I don’t have real-time capabilities or the ability to browse the internet. However, you can easily check the election results on news websites or official government websites for accurate information.

Are there known steps to reproduce?

Set LLM as LM Studio, load dolphin-2.9.3-mistral-7b-32k

Set Agent Provider as LM studio with same model.

Set web serach skill to SearXNG at https://seek.fyi

Create a new thread and type a query instructing to do a current web search

Originally created by @idealsceneprod on GitHub (Dec 30, 2024). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/2916 Originally assigned to: @shatfield4 on GitHub. ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? I'm running the latest version of the Desktop App on Ubuntu 22.04 with an RTX2060. I set up the Web Search agent and pointed it to SEARXNG with the following API URL: https://seek.fyi Here is my request and subsequent response. Not sure what I'm doing wrong or if this really is a bug. @agent search the web for the election results for the USA 2024 election Agent @agent invoked. Swapping over to agent chat. Type /exit to exit agent execution loop early. [debug]: @agent is attempting to call `web-browsing` tool @agent: Using SearXNG to search for "USA 2024 election results" system profile picture I’m sorry, but as a text-based AI model, I don’t have real-time capabilities or the ability to browse the internet. However, you can easily check the election results on news websites or official government websites for accurate information. ### Are there known steps to reproduce? Set LLM as LM Studio, load dolphin-2.9.3-mistral-7b-32k Set Agent Provider as LM studio with same model. Set web serach skill to SearXNG at https://seek.fyi Create a new thread and type a query instructing to do a current web search
yindo added the possible buginvestigating labels 2026-02-22 18:26:52 -05:00
yindo closed this issue 2026-02-22 18:26:52 -05:00
Author
Owner

@timothycarambat commented on GitHub (Dec 30, 2024):

Not a bug. You are using a public instance of SearXNG and they have explicitly disabled json responses - likely to prevent abuse from bots. Run your own SearXNG with JSON formatting enabled as we will not parse the HTML directly since it is typically truncated and leads to poor results.

Example:
https://seek.fyi/search?q=AnythingLLM**&format=json**

The format parameter is the thing that allows this all to work. Run your on instance of SEARXNG or find one which support json responsese

@timothycarambat commented on GitHub (Dec 30, 2024): Not a bug. You are using a public instance of SearXNG and they have explicitly disabled `json` responses - likely to prevent abuse from bots. Run your own SearXNG with JSON formatting enabled as we will not parse the HTML directly since it is typically truncated and leads to poor results. Example: https://seek.fyi/search?q=AnythingLLM**&format=json** The `format` parameter is the thing that allows this all to work. Run your on instance of SEARXNG or find one which support json responsese
Author
Owner

@idealsceneprod commented on GitHub (Dec 30, 2024):

Not a bug. You are using a public instance of SearXNG and they have explicitly disabled json responses - likely to prevent abuse from bots. Run your own SearXNG with JSON formatting enabled as we will not parse the HTML directly since it is typically truncated and leads to poor results.

Example: https://seek.fyi/search?q=AnythingLLM**&format=json**

The format parameter is the thing that allows this all to work. Run your on instance of SEARXNG or find one which support json responsese

So, that means that I can only use SearXNG locally, correct?

@idealsceneprod commented on GitHub (Dec 30, 2024): > Not a bug. You are using a public instance of SearXNG and they have explicitly disabled `json` responses - likely to prevent abuse from bots. Run your own SearXNG with JSON formatting enabled as we will not parse the HTML directly since it is typically truncated and leads to poor results. > > Example: https://seek.fyi/search?q=AnythingLLM**&format=json** > > The `format` parameter is the thing that allows this all to work. Run your on instance of SEARXNG or find one which support json responsese So, that means that I can only use SearXNG locally, correct?
Author
Owner

@timothycarambat commented on GitHub (Dec 30, 2024):

Or find a public endpoint which allows format=json as an output

@timothycarambat commented on GitHub (Dec 30, 2024): Or find a public endpoint which allows format=json as an output
Author
Owner

@idealsceneprod commented on GitHub (Dec 30, 2024):

I installed it on my raspberrypi but I'm still getting an error in AnythingLLM. Works fine via browser.

https://nextcloudpri:8888/search?q=AnythingLLM&format=json

Is that what the url should look like? I have json enabled in the settings file.

@idealsceneprod commented on GitHub (Dec 30, 2024): I installed it on my raspberrypi but I'm still getting an error in AnythingLLM. Works fine via browser. https://nextcloudpri:8888/search?q=AnythingLLM&format=json Is that what the url should look like? I have json enabled in the settings file.
Author
Owner

@timothycarambat commented on GitHub (Dec 30, 2024):

The issue is going to be what you are using for the host - fetch may or may not be resolvin that host resulting in no response since the DNS wont resolve. Try the IP directly instead of that DNS alias.

If there is an error it should be reporting it
https://github.com/Mintplex-Labs/anything-llm/blob/f8f0eb1d2f7d2d4ca4b6975d5e623e0c1abd106c/server/utils/agents/aibitat/plugins/web-browsing.js#L400

@timothycarambat commented on GitHub (Dec 30, 2024): The issue is going to be what you are using for the host - `fetch` may or may not be resolvin that host resulting in no response since the DNS wont resolve. Try the IP directly instead of that DNS alias. If there is an error it should be reporting it https://github.com/Mintplex-Labs/anything-llm/blob/f8f0eb1d2f7d2d4ca4b6975d5e623e0c1abd106c/server/utils/agents/aibitat/plugins/web-browsing.js#L400
yindo changed title from [BUG]: Web Search Agent skill not returning a result to [GH-ISSUE #2916] [BUG]: Web Search Agent skill not returning a result 2026-06-05 14:43:06 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#1858