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

# Knowledge base

> Gather documents so the AI can use them as reference, making answers more precise and contextual.

## What it is

The **Knowledge Base** lets you gather documents (manuals, policies, spreadsheets, texts, etc.) so the AI can use them as **reference**. When you connect a base to an agent or attach it to a conversation, answers become more **precise and aligned** to your context.

## How to access

* Sidebar → **Knowledge**
* To create: **New → Knowledge** button or the creation button on the page itself.

<Note>
  The page only appears if the **Workspace** module and the **Knowledge** permission are enabled for you.
</Note>

## Create a knowledge base

<Steps>
  <Step title="Click Create a knowledge base" />

  <Step title="Fill in the fields">
    * **Name** *(required)* — between **3 and 60 characters**.
    * **Description** *(optional)* — summary of the base's purpose.
  </Step>

  <Step title="Save">
    Click **Save**. A message confirms creation and you are taken to the base page to **add files**.
  </Step>
</Steps>

## Add files

On the base page, you can:

* **Drag and drop** files onto the indicated area, or
* Use the **add file** button to select from your computer, or
* **Add text content** directly (paste/write text that becomes a base document).

**Supported formats** include (among others):

```
pdf, txt, doc, docx, xls, xlsx, ppt, pptx, csv, tsv, json, xml, html, md, rtf,
odt, ods, odp, compressed files (zip, tar, gz, rar, 7z), and various code formats
(js, ts, py, java, sql, yaml, etc.)
```

<Note>
  Each file goes through processing (text extraction/indexing) before becoming available for use.
</Note>

## Manage the base

On the base page you can also:

* **Edit** name and description.
* **Update/reprocess (sync)** the content.
* **Remove** files individually.
* **Share** the base with users and groups (access control).

## Use the knowledge base

* **In an agent:** in the agent's **Knowledge** block, connect the base (see [Agents](/en/guia/agentes)).
* **In chat:** type `#` in the message or use **+ → Attach knowledge** and select the base.

## Best practices

* Give **descriptive names** to bases (e.g., "HR Policies 2026") to find them easily.
* Keep documents **up to date**; remove old versions to avoid outdated answers.
* Group documents by **topic** in separate bases, rather than one large generic base.
* Prefer documents with **selectable text** (searchable PDFs) for better extraction.

## Frequently asked questions

<AccordionGroup>
  <Accordion title="What is the file size/limit?">
    The size limit and accepted types follow the administrator's configuration (defined in **Admin Panel → Documents**). The formats listed above are those supported by the interface.
  </Accordion>

  <Accordion title="Why does my file appear as &#x22;processing&#x22;?">
    After upload, the system extracts and indexes the content. Wait for it to finish; only then can the base be used by the AI.
  </Accordion>

  <Accordion title="Can I use the same base in multiple agents?">
    Yes. A base can be connected to as many agents as you want.
  </Accordion>
</AccordionGroup>

## Known limitations

* The base name must be between **3 and 60 characters**.
* File types outside the supported list are not accepted.
* Size limits and processing parameters depend on the settings defined by the administrator.
* Creating/editing bases depends on the **Knowledge** permission in the Workspace.