# Overview

> Ground an agent's answers in documents synced from your existing tools.

Source: https://docs.runbear.io/agents/knowledge/overview

Last updated: 2026-10-07

Knowledge Base lets your agent give grounded, up-to-date answers by syncing documents from your existing tools. Instead of relying solely on its training data, the agent retrieves relevant content from your connected sources before responding — a technique known as Retrieval-Augmented Generation (RAG).

## How it works

When you connect a knowledge source, Runbear syncs the documents and indexes them for retrieval. During a conversation, the agent searches the indexed content, finds relevant passages, and uses them to formulate its response.

Sync is **incremental** — Runbear periodically checks for changes and updates only the content that has been added, modified, or removed.

## Knowledge Base vs Tool

Knowledge Base and [Tools](/agents/tools/overview.md) serve complementary purposes:

|                  | Knowledge Base                                             | Tool (MCP)                                        |
| ---------------- | ---------------------------------------------------------- | ------------------------------------------------- |
| **How it works** | Pre-synced documents searched at query time                | Real-time API calls on demand                     |
| **Best for**     | Answering questions from existing docs                     | Live lookups, actions, and writes                 |
| **Example**      | "What's our vacation policy?" → retrieves from the HR wiki | "Create a ticket for this bug" → calls Linear API |

Most agents benefit from both — connect a knowledge base for grounded answers and tools for real-time actions.

## Supported sources

| Source                                                | Description                                               |
| ----------------------------------------------------- | --------------------------------------------------------- |
| [**Notion**](/agents/knowledge/notion.md)             | Sync pages and databases from your Notion workspace       |
| [**Confluence**](/agents/knowledge/confluence.md)     | Sync spaces and pages from Atlassian Confluence           |
| [**Google Drive**](/agents/knowledge/google-drive.md) | Sync documents, spreadsheets, and files from Google Drive |
| [**Slack**](/agents/knowledge/slack.md)               | Sync messages and threads from selected Slack channels    |
| [**SharePoint**](/agents/knowledge/sharepoint.md)     | Sync documents and pages from Microsoft SharePoint        |
| [**Website**](/agents/knowledge/website.md)           | Crawl and sync content from any public website            |
| [**Local Files**](/agents/knowledge/local-files.md)   | Upload files directly from your computer                  |

## Supported file formats

For file-based sources — [Google Drive](/agents/knowledge/google-drive.md), [SharePoint](/agents/knowledge/sharepoint.md), and [Local Files](/agents/knowledge/local-files.md) — only files in a format Runbear can index are synced.

| Category      | Formats                                                                                |
| ------------- | -------------------------------------------------------------------------------------- |
| Documents     | `.pdf`, `.doc`, `.docx`                                                                |
| Presentations | `.pptx`                                                                                |
| Spreadsheets  | `.csv`, `.xls`, `.xlsx`                                                                |
| Text and data | `.txt`, `.md`, `.html`, `.htm`, `.json`, `.tex`, `.yaml`                               |
| Code          | `.c`, `.cpp`, `.cs`, `.css`, `.go`, `.java`, `.js`, `.php`, `.py`, `.rb`, `.sh`, `.ts` |

Spreadsheet, CSV, and YAML files are converted to text when indexed; other formats are indexed as they are. `.yaml` can only be synced from Google Drive — local uploads don't accept it.

Google Docs, Sheets, and Slides have no file to download, so Runbear exports them automatically: Docs become Markdown, Sheets become `.xlsx`, and Slides become `.pptx`.

For Google Drive and SharePoint, a file in any other format is not dropped silently. The file picker marks it **Unsupported**, and the source lists it in a skipped-files notice — convert the file to a supported format and it syncs on the next run.

## Setup

1. Open your agent in the Runbear dashboard and go to the **Knowledge** tab.
2. Select a source.
3. Authorize access to the source (for cloud-based sources like Notion or Google Drive).
4. Select the specific content to sync — pages, folders, channels, or URLs depending on the source.
5. Runbear begins the initial sync. You can check progress on the sync status screen.

Once the initial sync completes, the agent can immediately use the content to answer questions.

For detailed instructions, see the guide for each source above.

> **Connected through another member's account**
>
> Notion, Confluence, SharePoint, and Google Drive (OAuth) knowledge bases are connected through the account of the member who set them up. Another member who opens the source can connect their own account instead, which moves the knowledge base to their account and keeps everything already synced. The one exception is a Confluence knowledge base connected with an API token: only the member who connected it can change it, and for everyone else its pill is disabled.

## Keeping content up to date

After the initial sync, Runbear periodically checks for changes and updates the index automatically. There is no manual refresh needed — your agent always has access to the latest content.