You need to make your Copilot agent answer questions using content stored in SharePoint document libraries. Copilot Studio can pull data from SharePoint sites to ground responses in your organization’s files. This article explains how to connect SharePoint as a knowledge source in Copilot Studio. You will learn the prerequisites, the exact configuration steps, and what to avoid when setting up the connection.
Key Takeaways: Connecting SharePoint to Copilot Studio
- Copilot Studio agent > Topics > Knowledge sources > Add knowledge source > SharePoint: The exact menu path to start the connection wizard.
- SharePoint site URL or document library URL: You must provide a specific URL rather than relying on auto-discovery.
- Permissions and authentication: The agent uses the maker’s credentials by default; adjust to end-user authentication for production use.
What SharePoint as a Knowledge Source Means in Copilot Studio
Copilot Studio lets you build custom agents that answer questions using data from Microsoft 365 sources. SharePoint is one of the supported knowledge sources. When you connect a SharePoint site or document library, the agent indexes the content from supported file types. The agent then uses that indexed content to generate answers to user queries.
The connection works through the Microsoft Graph API. Copilot Studio reads files from the specified SharePoint location. Supported file types include Word documents, PowerPoint presentations, PDFs, Excel spreadsheets, and text files. The agent does not store a copy of the files. It indexes the text content and uses it only at query time.
Prerequisites for Connecting SharePoint
Before you start, confirm the following requirements are met:
- You have a Copilot Studio license with the custom agent capability. The license is included with Copilot for Microsoft 365 or as a standalone add-on.
- You have at least Contributor access to the SharePoint site or document library you want to connect.
- The SharePoint site is a modern team site or communication site. Classic sites may not be fully supported.
- Files in the SharePoint location are in a supported format. Unsupported formats like ZIP, images without text, or executable files are ignored.
Steps to Connect SharePoint as a Knowledge Source in Copilot Studio
Follow these steps to add SharePoint content to your Copilot agent. The steps assume you already have a Copilot Studio agent created. If not, create a new agent first from the Copilot Studio home page.
- Open your agent in Copilot Studio
Sign in to Copilot Studio at copilotstudio.microsoft.com. On the left navigation pane, select Agents. Click the name of the agent you want to modify. The agent overview page opens. - Go to the Topics section
On the agent overview page, select the Topics tab from the top menu. This is where you manage the knowledge sources and conversation flows for your agent. - Open the knowledge source settings
On the Topics page, look for the Knowledge sources section. Click Add knowledge source. A dropdown menu appears with a list of available source types. - Select SharePoint as the source type
From the dropdown, choose SharePoint. A new panel opens on the right side of the screen. - Enter the SharePoint URL
In the URL field, paste the full URL of the SharePoint site or document library you want to connect. For a site-level connection, use the site URL such ashttps://contoso.sharepoint.com/sites/Projects. For a library-level connection, use the library URL such ashttps://contoso.sharepoint.com/sites/Projects/Shared%20Documents. Click Add. - Confirm the connection
Copilot Studio validates the URL and your permissions. If the connection succeeds, the SharePoint source appears in the list of knowledge sources. The agent begins indexing the content. Indexing time depends on the number and size of files. You can continue editing other parts of the agent while indexing runs. - Test the knowledge source
After indexing completes, open the Test your agent pane on the right side of the Topics page. Type a question that relates to the content in your connected SharePoint library. The agent should return an answer based on that content. If the answer is generic or incorrect, verify that the files are indexed and that the query is relevant.
Common Mistakes and Things to Avoid When Connecting SharePoint
Agent returns no answer from SharePoint content
If the agent does not use the SharePoint content, check the indexing status. In the knowledge sources list, look for a status indicator next to the SharePoint source. A green check mark means indexing is complete. A yellow warning or red error means indexing failed or is still running. Re-add the source if the status remains in error after 24 hours.
Using a personal OneDrive URL instead of a SharePoint URL
Copilot Studio accepts only SharePoint site or document library URLs. OneDrive for Business URLs are not supported. If you try to add a OneDrive URL, the system shows a validation error. Copy the correct SharePoint URL from your browser address bar when you are on the SharePoint site or library page.
Files are not indexed despite correct URL
Some file types are not indexed by Copilot Studio. Unsupported types include image-only PDFs, scanned documents without OCR, ZIP archives, and executable files. Convert unsupported files to a supported format such as Word or text before uploading them to SharePoint. The agent also skips files larger than 100 MB. Split large files into smaller parts if needed.
Authentication fails for end users
By default, the SharePoint knowledge source uses the credentials of the person who added the source. When other users ask questions, the agent may not have permission to read the files. To fix this, configure the agent to use end-user authentication. In the agent settings, under Security, select Authenticate with the user who is chatting. This ensures each user sees only content they have permission to access.
SharePoint Knowledge Source vs Other Copilot Studio Knowledge Sources
| Item | SharePoint | Other Sources (Web, Dataverse, Custom) |
|---|---|---|
| Data location | Specific SharePoint site or document library | Public web URLs, Dataverse tables, or uploaded files |
| Authentication | Maker credentials or end-user authentication | Varies by source: public web requires no auth, Dataverse uses environment roles |
| Supported file types | Word, PowerPoint, PDF, Excel, text files | Web sources: HTML content only; Dataverse: table rows; Custom: uploaded documents |
| Update frequency | Index is refreshed every 24 hours or on manual re-sync | Web sources re-index on schedule; Dataverse uses real-time queries |
| Best use case | Internal documentation, project files, policy documents | Public knowledge bases, structured data, proprietary content |
You can now connect SharePoint as a knowledge source in Copilot Studio and control exactly which documents your agent uses. After the connection is live, test the agent with several real queries to verify the answers are accurate. For production agents, switch to end-user authentication so permissions are respected per person. Consider adding multiple SharePoint libraries or combining SharePoint with a web source if your content spans both internal and public data.