> ## Documentation Index
> Fetch the complete documentation index at: https://docs.anygen.io/llms.txt
> Use this file to discover all available pages before exploring further.

# AnyGen Knowledge Base · Extract from Link

> Import web pages or Feishu / Lark cloud documents into your AI assistant's knowledge base.

<Info>
  This guide explains how to use the **Extract from Link** feature to import web pages or Feishu / Lark cloud documents into your AI assistant's knowledge base.
</Info>

<Tip>
  **Quick Summary**: Copy a Feishu document or wiki link → paste it into the **Extract from Link** dialog → click **Add**. The system will automatically crawl and add the content so the AI can reference it.
</Tip>

## Entry Point

On the AI assistant's **Knowledge Base** edit page, click **Extract from Link**. A dialog appears:

* **Title**: Extract from Link
* **Supported**: web links, Feishu / Lark cloud document links
* **Input**: paste the target link
* **Hint at bottom**: Feishu / Lark wiki links extract up to 1000 documents

## Knowledge Base Capacity Limits

Per-assistant knowledge base limits (apply to both **Extract from Link** and **Upload File**):

| Dimension                                                     | Limit                |
| ------------------------------------------------------------- | -------------------- |
| Single Feishu wiki global link (`/wiki/space/...`) extraction | Up to 1000 documents |
| Total files per assistant knowledge base                      | Up to 2000           |
| Total size per assistant knowledge base                       | Up to 1 GB           |

## Supported Link Types

| Type                                  | Example                                | Description                                   |
| ------------------------------------- | -------------------------------------- | --------------------------------------------- |
| Web link                              | `https://example.com/article`          | Any publicly accessible HTTP/HTTPS page       |
| Feishu cloud doc (single)             | `https://xxx.feishu.cn/docx/xxxx`      | Crawls the full document                      |
| Feishu Base                           | `https://xxx.feishu.cn/base/xxxx`      | Crawls the Base content                       |
| Feishu Sheet                          | `https://xxx.feishu.cn/sheets/xxxx`    | Crawls the Sheet content                      |
| Feishu wiki node                      | `https://xxx.feishu.cn/wiki/xxxx`      | Crawls only this single node (no descendants) |
| Feishu wiki global link               | `https://my.feishu.cn/wiki/space/xxxx` | Crawls all documents in the wiki (up to 1000) |
| Lark doc / Sheet / Base / Wiki / file | `https://xxx.larksuite.com/docx/...`   | Same as above, overseas environment           |

## Unsupported Content

<Warning>
  The following link types **cannot** be imported; please use file upload or another method:

  * **WeChat Public Account articles** (`mp.weixin.qq.com`): not supported by the crawler; blocked on submission
  * **Feishu / Lark** **Slides**: only **native** Slides created in Feishu / Lark (links like`slides/...`) are unsupported

  To import a native Slide created in Feishu / Lark, either copy its content to a regular Feishu / Lark cloud doc (docx) first, or export it as a local file (`.pptx`) and upload it into Feishu / Lark, then import via its file link.
</Warning>

## How to Import a Single Feishu / Lark Cloud Document

<Steps>
  <Step title="Open the document">
    Open the target Feishu / Lark native document or uploaded file cloud document in your browser.
  </Step>

  <Step title="Copy the URL">
    Copy the full URL from the address bar (e.g. `https://xxx.feishu.cn/docx/...` for a native doc, or `https://xxx.feishu.cn/file/...` for a file cloud document).
  </Step>

  <Step title="Open Extract from Link">
    In AnyGen, open the AI assistant's **Knowledge Base** → click **Extract from Link**.
  </Step>

  <Step title="Paste the link">
    Paste the link into the input box.
  </Step>

  <Step title="Click Add">
    Click **Add**. The content is added to the assistant's knowledge; the AI will reference it automatically.
  </Step>
</Steps>

## How to Import a Feishu / Lark Wiki

Two methods, choose based on your use case:

### Method 1: Wiki Node Link Import (single node)

Use this when you want to import a single wiki node as a knowledge document.

<Steps>
  <Step title="Open the wiki">
    Open the target wiki in Feishu.
  </Step>

  <Step title="Find the node">
    Find the node you want to import.
  </Step>

  <Step title="Copy the node link">
    Copy that node's link (e.g. `https://xxx.feishu.cn/wiki/...`).
  </Step>

  <Step title="Paste and add">
    Paste it into the **Extract from Link** dialog, click **Add**.
  </Step>

  <Step title="Single node only">
    The system crawls **only this node** — it does **not** recurse into descendants. To import the entire wiki, use Method 2.
  </Step>
</Steps>

### Method 2: Global Wiki Import (up to 1000)

Use this when you want to import an entire wiki from the root in one go.

<Steps>
  <Step title="Open the wiki">
    Open the target wiki in Feishu.
  </Step>

  <Step title="Click Share Wiki">
    Click **Share Wiki** in the lower-left.

    <Frame>
      <img src="https://mintcdn.com/anygen/0B2yxSfoR7ciQpkc/images/20260529-111224.jpeg?fit=max&auto=format&n=0B2yxSfoR7ciQpkc&q=85&s=49e0c5bb35623bc544c6ba51e79fbbde" alt="20260529 111224" width="2758" height="1538" data-path="images/20260529-111224.jpeg" />
    </Frame>
  </Step>

  <Step title="Copy the global share link">
    Copy the generated global share link (e.g. `https://my.feishu.cn/wiki/space/...`).

    <Frame>
      <img src="https://mintcdn.com/anygen/0B2yxSfoR7ciQpkc/images/20260529-111237.jpeg?fit=max&auto=format&n=0B2yxSfoR7ciQpkc&q=85&s=69189b1c2caaacb90eb742954a0dc56d" alt="20260529 111237" width="2740" height="1538" data-path="images/20260529-111237.jpeg" />
    </Frame>
  </Step>

  <Step title="Paste and add">
    Paste it into **Extract from Link**, click **Add**.
  </Step>

  <Step title="Bulk crawl">
    The system detects it as a global wiki link and crawls all documents, **up to 1000**.
  </Step>
</Steps>

## Permissions (Important)

<Warning>
  Access rules differ by link type:

  * **Feishu document / Base / wiki**: access **inherits from the current signed-in user's Feishu read permission**. If you can open the link in Feishu, AnyGen can crawl it — **no public-link setting is required**.
  * **Generic web page**: must be **publicly accessible without login**. Pages that require login, are behind a paywall, or are intranet-only cannot be crawled.

  Import failures are usually caused by one of these conditions not being met.
</Warning>

### Recommended Steps

1. Confirm the Feishu account currently signed in to AnyGen can open the target document / wiki in Feishu.
2. If you can't open it, request read access from the document / wiki owner.
3. Once you can open it, copy the link and paste it into the **Extract from Link** dialog.

<Tip>
  **Wiki child node permissions**: when bulk-importing via the global link (Method 2), check in **Wiki Settings → Permission Management** that all child nodes you need imported are readable by the current user. If only some sub-documents are crawled, it is usually because child nodes did not inherit the permission.
</Tip>

## FAQ

<AccordionGroup>
  <Accordion title="Q1: After pasting, it says &#x22;Please enter a valid link&#x22;">
    The link doesn't match URL format. Check:

    * The URL is copied in full from the browser address bar
    * Starts with `http://` or `https://`
    * No stray punctuation or whitespace
  </Accordion>

  <Accordion title="Q2: Submission fails with a fetch error">
    The most common cause is **insufficient permission**: Feishu content requires the current signed-in user to have Feishu read access; generic web pages must be openable without login.

    Investigation order:

    1. **Can you open it in Feishu?** Open the link in your browser with the same Feishu account currently signed in to AnyGen and confirm you can read it.
    2. **Child-node permission** (global link / Method 2 only): see the wiki child-node permission note under **Permissions** above.
    3. **Link type**: confirm it's not a WeChat article, or a **native** Feishu / Lark Slide (created in Feishu / Lark).
    4. **Network reachability**: for web links, confirm the page is reachable on the public Internet (not an intranet-only domain).
  </Accordion>

  <Accordion title="Q3: My wiki has more than 1000 documents but only 1000 were imported">
    A global share link (`my.feishu.cn/wiki/space/...`) imports up to 1000 documents per call. If your wiki exceeds 1000 documents, suggestions:

    * Split the content into multiple sub-wikis, generate a global share link for each, and import them one by one.
    * Or pick the 1000 most-referenced documents, organize them into a new wiki, and import that.
  </Accordion>

  <Accordion title="Q4: A web page is only partially crawled">
    Pages that rely on JavaScript rendering (SPAs, login-required, paywalled) are unreliable. Suggestions:

    * Copy the content into a Feishu cloud doc first, then import via the Feishu link.
    * Or directly upload a PDF / Word / Markdown file.
  </Accordion>

  <Accordion title="Q5: Does AnyGen auto-sync after the source Feishu Doc/Sheet/Base updates?">
    Feishu Docs, Sheets, and Base all support an **Auto Sync** option. On the assistant's **Knowledge Base** page, find the **Auto Sync** config and turn it on — the system will periodically re-crawl the latest content.
  </Accordion>

  <Accordion title="Q6: Can I import a PPT / Excel / PDF that I uploaded to Feishu / Lark?">
    Yes. An uploaded file is stored as a **file cloud document** (`/file/...` link) and can be imported via Extract from Link. This differs from a **native cloud document**  Slide created directly in Feishu / Lark, which is **not** supported — see **Unsupported Content** above.
  </Accordion>
</AccordionGroup>

## When to use "Extract from Link" vs Upload File

| Scenario                                                   | Recommended                                                  |   |
| ---------------------------------------------------------- | ------------------------------------------------------------ | - |
| Content updates continuously (operations manual, API docs) | **Extract from Link** (Feishu doc + Auto Sync)               |   |
| One-time reference material (contract, report, PDF)        | **Upload File**                                              |   |
| Intranet / paywalled / dynamically-rendered pages          | **Upload File** (paste into a Feishu doc first, then import) |   |

## Feedback

If you encounter an issue not covered here or have suggestions, please contact the AnyGen team.