Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-18 19:34:33 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3318] [FEAT]: Agent Skill - Scrape Websites option to not summarise #2134

New Issue
Closed
opened 2026-02-22 18:28:17 -05:00 by yindo · 1 comment
yindo commented 2026-02-22 18:28:17 -05:00
Owner
Copy Link
Copy Source

Originally created by @CaffiendDevKit on GitHub (Feb 22, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3318

Originally assigned to: @shatfield4 on GitHub.

Version

Docker Hub - Latest - I guess v1.7.4

What would you like to see?

I am trying to follow the Agent Flows Tutorial but the web scraper or the agent always summarizes the HTML. This causes all the links to be the same and invalid.

It would be nice to have an explicit option to tell it to pass the raw scrapped content to the next block.

I even tried with the API Call block and the content was also summarized.

I attempted to alter the agent description to tell it to not summarize large responses but that didn't seem to make a difference.

Additionally I noticed that even when looking at a small return less than 100 tokens it would still attempt to summarize saying it was too large. So it appears to be a hard coded step.

The "summarized" output, note all links are iterated from item?id=123456 and other changes such as the user.

<!DOCTYPE html>
<html>
<head>
    <title>Hacker News</title>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <link rel="stylesheet" href="https://news.ycombinator.com/news.css">
</head>
<body>
    <table>
        <tr>
            <td>
                <span>1.</span>
                <a href="https://www.example.com">Example Title</a>
                <span>(123 points)</span>
                <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 1 hour ago</span>
                <span>| <a href="https://news.ycombinator.com/item?id=123456">123 comments</a></span>
            </td>
        </tr>
        <tr>
            <td>
                <span>2.</span>
                <a href="https://www.example.com">Deep Dive into Llama3 AI Model</a>
                <span>(90 points)</span>
                <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 2 hours ago</span>
                <span>| <a href="https://news.ycombinator.com/item?id=123457">90 comments</a></span>
            </td>
        </tr>
        <tr>
            <td>
                <span>3.</span>
                <a href="https://www.example.com">AI Accumulating Wealth</a>
                <span>(78 points)</span>
                <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 3 hours ago</span>
                <span>| <a href="https://news.ycombinator.com/item?id=123458">78 comments</a></span>
            </td>
        </tr>
        <!-- more table rows with submission information -->
    </table>
    <table>
        <tr>
            <td>
                <a href="https://news.ycombinator.com/more">more</a>
            </td>
        </tr>
    </table>
</body>
</html>
Originally created by @CaffiendDevKit on GitHub (Feb 22, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3318 Originally assigned to: @shatfield4 on GitHub. ### Version Docker Hub - Latest - I guess v1.7.4 ### What would you like to see? I am trying to follow the [Agent Flows Tutorial ](https://docs.anythingllm.com/agent-flows/tutorial-hackernews) but the web scraper or the agent always summarizes the HTML. This causes all the links to be the same and invalid. It would be nice to have an explicit option to tell it to pass the raw scrapped content to the next block. I even tried with the API Call block and the content was also summarized. I attempted to alter the agent description to tell it to not summarize large responses but that didn't seem to make a difference. Additionally I noticed that even when looking at a small return less than 100 tokens it would still attempt to summarize saying it was too large. So it appears to be a hard coded step. The "summarized" output, note all links are iterated from `item?id=123456` and other changes such as the user. ```html <!DOCTYPE html> <html> <head> <title>Hacker News</title> <meta charset="utf-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <link rel="stylesheet" href="https://news.ycombinator.com/news.css"> </head> <body> <table> <tr> <td> <span>1.</span> <a href="https://www.example.com">Example Title</a> <span>(123 points)</span> <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 1 hour ago</span> <span>| <a href="https://news.ycombinator.com/item?id=123456">123 comments</a></span> </td> </tr> <tr> <td> <span>2.</span> <a href="https://www.example.com">Deep Dive into Llama3 AI Model</a> <span>(90 points)</span> <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 2 hours ago</span> <span>| <a href="https://news.ycombinator.com/item?id=123457">90 comments</a></span> </td> </tr> <tr> <td> <span>3.</span> <a href="https://www.example.com">AI Accumulating Wealth</a> <span>(78 points)</span> <span>by <a href="https://news.ycombinator.com/user?id=exampleuser">exampleuser</a> 3 hours ago</span> <span>| <a href="https://news.ycombinator.com/item?id=123458">78 comments</a></span> </td> </tr> <!-- more table rows with submission information --> </table> <table> <tr> <td> <a href="https://news.ycombinator.com/more">more</a> </td> </tr> </table> </body> </html> ```
yindo added the enhancementfeature request labels 2026-02-22 18:28:17 -05:00
yindo closed this issue 2026-02-22 18:28:17 -05:00
yindo commented 2026-02-22 18:28:17 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Feb 22, 2025):

This is a good point, the tool should not interpret the text, even if large, since that is out of scope for the block and the data should be handed off the next block as raw output based on findings.

@timothycarambat commented on GitHub (Feb 22, 2025): This is a good point, the tool should not interpret the text, even if large, since that is out of scope for the block and the data should be handed off the next block as raw output based on findings.
yindo changed title from [FEAT]: Agent Skill - Scrape Websites option to not summarise to [GH-ISSUE #3318] [FEAT]: Agent Skill - Scrape Websites option to not summarise 2026-06-05 14:44:45 -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#2134
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?