Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-21 01:25:28 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #4154] [FEAT]: Agent functionality in embed widget #2642

New Issue
Open
opened 2026-02-22 18:30:34 -05:00 by yindo · 7 comments
yindo commented 2026-02-22 18:30:34 -05:00
Owner
Copy Link
Copy Source

Originally created by @Alminc91 on GitHub (Jul 15, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4154

What would you like to see?

Hello.

How difficult would it be to reference or use the agent functionality from the web app in the embed widget? For example, summarize the current conversation and send an email to a specific customer support mail. Or call API-endpoints for actions.

Would that be a very big effort? I saw one older ticket requesting multiple features for the embed widget, including this one. Thanks!

Originally created by @Alminc91 on GitHub (Jul 15, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/4154 ### What would you like to see? Hello. How difficult would it be to reference or use the agent functionality from the web app in the embed widget? For example, summarize the current conversation and send an email to a specific customer support mail. Or call API-endpoints for actions. Would that be a very big effort? I saw one older ticket requesting multiple features for the embed widget, including this one. Thanks!
yindo added the enhancementfeature request labels 2026-02-22 18:30:34 -05:00
yindo commented 2026-02-22 18:30:34 -05:00
Author
Owner
Copy Link
Copy Source

@TadasPik commented on GitHub (Jul 16, 2025):

+1 on this. Also, without a need to explicitly invoke @agent.
Just an idea. plugin.json could have smth among the lines of "invocation_mode": "auto", or "invocation_mode": "agent" by default to not alter existing behavior.

@TadasPik commented on GitHub (Jul 16, 2025): +1 on this. Also, without a need to explicitly invoke @agent. Just an idea. plugin.json could have smth among the lines of "invocation_mode": "auto", or "invocation_mode": "agent" by default to not alter existing behavior.
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@Alminc91 commented on GitHub (Jul 16, 2025):

Edit: I see that @TadasPik had the same thought! That would be amazing! I saw that in Open WebUI or also Coding IDE's of course, that automatically can call multiple tools from natural language.

I am also thinking: is it possible to call tools naturally? I did not succeed with pasting an url and saying to search for something in that url. If I used @agent, it worked.

Maybe that is another feature request, but that would make (for the embed widget) conversations more natural. Users or customers won't call @agent for a tool, but ask in natural language. Or is this already possible? Thanks!

That would be a great combo. I think an estimate and feasibility for both features would be a great first step.

@Alminc91 commented on GitHub (Jul 16, 2025): Edit: I see that @TadasPik had the same thought! That would be amazing! I saw that in Open WebUI or also Coding IDE's of course, that automatically can call multiple tools from natural language. I am also thinking: is it possible to call tools naturally? I did not succeed with pasting an url and saying to search for something in that url. If I used @agent, it worked. Maybe that is another feature request, but that would make (for the embed widget) conversations more natural. Users or customers won't call @agent for a tool, but ask in natural language. Or is this already possible? Thanks! That would be a great combo. I think an estimate and feasibility for both features would be a great first step.
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Jul 16, 2025):

understood 👍

@timothycarambat commented on GitHub (Jul 16, 2025): understood 👍
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@TadasPik commented on GitHub (Aug 18, 2025):

Any plans / timelines for this?

@TadasPik commented on GitHub (Aug 18, 2025): Any plans / timelines for this?
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Aug 18, 2025):

The main reason we havent done this is simply because it can inadvertently expose a user to unknown risks. The whole conversation around MCPs being too powerful I can imagine someone has them in a chat widget and a malicious, unknown untrustworthy user starts using the skills over an over either wasting tokens or making calls to tools trying to accomplish some nefarious task.

When ever user needs an account through an admin you can at least anticipate or approve trustworthy people. That chat widget, not so much

@timothycarambat commented on GitHub (Aug 18, 2025): The main reason we havent done this is simply because it can inadvertently expose a user to unknown risks. The whole conversation around MCPs being too powerful I can imagine someone has them in a chat widget and a malicious, unknown untrustworthy user starts using the skills over an over either wasting tokens or making calls to tools trying to accomplish some nefarious task. When ever user needs an account through an admin you can at least anticipate or approve trustworthy people. That chat widget, not so much
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@TadasPik commented on GitHub (Aug 18, 2025):

Thought of that as well. But maybe here is another way looking at it. Lets take following example.

I, as a anythingllm user, create a tool to get notified if user wants to inquiry something.
I add a tool to send me an email over api that user is interested in my services.
In the tool i use 2 apis - validate email, and then send email.
In the tool i add invocation_mode: auto - thus allowing this tool to be executed w/o @agent directive (from embedded chat)

Other tools wont be touched, since by default, invocation_mode is agent

I'll get my email, and won't waste any tokens as far as i can tell no?

Maybe there is another way to achieve all this?

@TadasPik commented on GitHub (Aug 18, 2025): Thought of that as well. But maybe here is another way looking at it. Lets take following example. I, as a anythingllm user, create a tool to get notified if user wants to inquiry something. I add a tool to send me an email over api that user is interested in my services. In the tool i use 2 apis - validate email, and then send email. In the tool i add invocation_mode: auto - thus allowing this tool to be executed w/o @agent directive (from embedded chat) Other tools wont be touched, since by default, invocation_mode is agent I'll get my email, and won't waste any tokens as far as i can tell no? Maybe there is another way to achieve all this?
yindo commented 2026-02-22 18:30:35 -05:00
Author
Owner
Copy Link
Copy Source

@EfrosIonelu commented on GitHub (Aug 29, 2025):

@TadasPik you can use https://github.com/Mintplex-Labs/anythingllm-embed/pull/12

@EfrosIonelu commented on GitHub (Aug 29, 2025): @TadasPik you can use https://github.com/Mintplex-Labs/anythingllm-embed/pull/12
yindo changed title from [FEAT]: Agent functionality in embed widget to [GH-ISSUE #4154] [FEAT]: Agent functionality in embed widget 2026-06-05 14:47:42 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label enhancement feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2642
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?