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

# Knowledge

> Give the agent facts to look up — pricing, policies, FAQs — instead of relying on what's baked into its instructions.

## What is this?

An agent with only [Instructions](https://wabis.mintlify.site/workspace/ai-agents/tabs/instructions.md) can chat nicely but doesn't know your facts. **Knowledge** is its reference binder — the things it looks up rather than the things it's told to memorize.

## What can I do here?

Click **Add Source** and choose a **Source Type**:

| Source Type | Use it when |
| - | - |
| **Manual Text** | You want to type or paste facts directly — prices, opening hours |
| **Website URL** | Your answers already live on a web page |
| **Sitemap Crawl** | You want the agent to read many pages from your site at once |
| **llms.txt Crawl** | Your site publishes an `llms.txt` guide for AI |
| **PDF, Word, Markdown, Text, CSV, Excel** | Your facts are in a document or spreadsheet |

Give it an optional **Title**, add the content/link/file, and save.

To add many sources at once, click **Bulk Import**. Choose a method:

| Method | What it does |
| - | - |
| **Upload Files** | Upload multiple PDF, Word (.docx), Markdown, Text, CSV, or Excel files in one go — each becomes its own source |
| **Import URLs** | Paste a list of URLs (one per line) — each URL becomes its own source |
| **Sitemap** | Enter a sitemap.xml URL and click **Detect Pages** — imports every page listed (up to the displayed limit) |
| **llms.txt** | Enter an llms.txt URL and click **Detect Pages** — imports every linked page |
| **URL List** | Upload a CSV, Excel, or text file, choose the column that holds the URLs — the file itself is not added as knowledge |

After confirming, all items are queued and processed in the background. A progress summary shows **Total**, **Processing**, **Completed**, **Failed**, and **Skipped** counts. You can close the dialog while processing continues, and use **Retry Failed** to re-queue any that errored.

## How to use it

### Wait for it to finish processing

Each source moves through a status: **Pending → Processing → Processed**. Wait for **Processed** before you test the agent against it. **Failed** means the link didn't load or the file was empty — fix that and add it again.

> [!TIP]
>
> Start small. One or two good sources — a price list and a short FAQ — beat a huge pile of old documents. Short, accurate, up-to-date facts produce better answers than volume.

### Pair it with a rule in Instructions

Knowledge only helps if the agent is told to actually rely on it. Add a line like *"Only use the facts in your knowledge. Never invent prices, stock, or policies."* to [Instructions](https://wabis.mintlify.site/workspace/ai-agents/tabs/instructions.md) — otherwise the model may fall back on a plausible-sounding guess.

## Troubleshooting / Technical Notes

* **A source shows Failed.** Check that the link opens in a normal browser, or that the file isn't empty or corrupted, then add it again.
* **The agent gives wrong or made-up answers.** Confirm the relevant source shows **Processed**, and tighten [Instructions](https://wabis.mintlify.site/workspace/ai-agents/tabs/instructions.md) to say it should only use knowledge it has.
* **The agent seems to ignore a source you just added.** Give it a moment to finish processing, then start a **New Conversation** in [Test](https://wabis.mintlify.site/workspace/ai-agents/tabs/test.md) — an in-progress test session may not pick up a source added mid-conversation.

## Related docs

* [Instructions](https://wabis.mintlify.site/workspace/ai-agents/tabs/instructions.md)
* [Capabilities](https://wabis.mintlify.site/workspace/ai-agents/tabs/capabilities.md)
* [Test](https://wabis.mintlify.site/workspace/ai-agents/tabs/test.md)


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