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

# Adding context to your knowledge base

> How to give your agent the information it needs to answer visitor questions well.

Open **Knowledge sources** under Agent in your dashboard to add and manage what your agent knows. Each source shows a Ready badge once it can be used, or Needs review if it is held back.

## Why the knowledge base matters

Your agent is only as good as the context you give it. Lark does not make things up. It answers from what you have added, and if the answer is not in your knowledge base it says so and offers to pass the visitor to you.

## Three ways to add context

### Paste text

Copy and paste your services list, pricing page, FAQs, team bios or process descriptions. Anything you would want a visitor to know.

### Upload a file

Upload text (.txt), Markdown (.md), HTML, JSON, PDF or Word (.docx) files. Lark reads the text and keeps it, then discards the original file. Older .doc files are not supported, so save them as .docx first.

### Add a website URL

Add a page from your website and Lark pulls in the text, up to 5 pages per source.

## Limits

Limits are measured on extracted text content, not file size. A large file that contains little readable text counts much less than its file size suggests.

| Plan | Per file | Total extracted text |
| - | - | - |
| Starter | 2 MB | 5 MB |
| Pro | 2 MB | 20 MB |
| Enterprise | Custom | Custom |

Sources with very little readable text, such as scanned PDFs, are held for review and are not used in answers until you approve them.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.