Agent Data Sources

The Slite Agent connects to the tools your team already uses — Slack, Linear, Notion, GitHub, Google Drive, and more — so you can search and get AI answers across your company's knowledge in one place.
This guide covers how sources work, how access is controlled, and how to manage connected tools. For setup steps on a specific source, click through from the list below.

Where to add your Agent Data Sources

How data sources work

Every agent source in the Slite Agent has two properties that shape how your team uses it: a setup method (how it connects) and an access model (who can query it once connected).

Setup methods

OAuth
Click connect, sign in to the source, grant Slite access. No admin configuration in the source itself.
Requires configuration in the source
Before the Slite Agent can connect, an admin needs to set something up inside the source first — typically installing an app, creating a service account, or configuring a delegation. The guide for each source walks through exactly what's needed.
No auth required
Content is either uploaded directly into the Slite Agent or is publicly accessible, so no external authentication is needed.

Access models

Shared
All workspace members can query the source. No account in the connected tool is required. Workspace admins can further restrict access per connection — see Controlling access to shared sources below.
Mirrored
The Slite Agent matches each member to their account in the connected tool by email. Members can only query content they already have access to in that tool. No matching email account = no results from that source.
The access model is set by the source type and can't be changed. You'll see it as a badge on each connection in the Agent Sources panel.


Supported sources

Click any source for setup steps and details on what's synced.
Tool
What we sync
Setup
Access
Tasks, comments, and metadata from selected projects (paginated via remote search API)
OAuth
Mirrored
Records from selected objects and lists
OAuth
Shared + Mirrored
Table descriptions from connected projects
Requires config in source
Mirrored
Pages (no attachments or comments)
OAuth
Mirrored
Custom data indexed via the public API
No auth required
Shared
.pdf, .pptx, .docx, .html, .txt
No auth required
Shared
All .md files in connected repos
No auth required
Shared
Issues and PRs from public and private repos (no comments or media)
OAuth
Shared
Issues and merge requests from public and private projects (no comments)
Requires config in source
Shared
Docs, Slides, .docx, .pptx, .pdf (no Sheets, images, or videos)
Requires config in source
Mirrored
1,000 most recent deals (max 100 properties each)
OAuth
Shared
Conversations (no articles, contacts, or notes)
OAuth
Mirrored
Issue descriptions (1 year) and metadata (no comments)
OAuth
Mirrored
Issues, comments, and metadata from synced teams
OAuth
Mirrored
Pages under selected roots (no database objects)
OAuth
Shared
Records from connected objects
Requires config in source
Mirrored
Pages and OneDrive files (.pdf, .pptx, .docx)
Requires config in source
Mirrored
Threads from public and private channels
OAuth
Mirrored
Your entire documentation
No auth required
Mirrored
Page content, metadata, and structure
No auth required
Shared
 Zendesk  (beta)
Tickets, including comments and replies, from the past year
Requires config in source
Mirrored
 Zoom  (beta)
Meeting transcripts
Requires config in source
Mirrored
Attio combines both access models depending on the content type — see its guide for details.

Sources and root nodes

Two different things are often both called a "source", and access works differently for each.
A source is the technical connection between Slite and another tool — one Slack workspace, one Notion workspace, one Confluence site. You set it up once.
Root nodes are the items inside that connection you have chosen to sync: individual Slack channels, Confluence spaces, Google Drive folders, Notion top-level pages. One source can hold many root nodes, and each one is managed independently.
In the Agent Sources panel, the left column lists sources. Selecting one shows its root nodes on the right, and it is the root nodes that carry an Access setting.
Some connections have only one root node, so the source and the root node look like the same thing. Notion works this way — the connected workspace is both. Connections such as Slack or Confluence make the two levels easier to see, because one connection holds many channels or spaces.

Controlling access to a root node

For Shared sources, workspace admins can set who in Slite can query each root node. Open the source in the Agent Sources panel, then adjust access on the root node:
  • Everyone at {workspace} — applies to all workspace members.
  • Specific users or user groups — add them individually and set each one.
The same three roles are available on every row:
  • Can manage — can query the item, and can change who else has access to it.
  • Can access — can query the item. It shows up in Slite Agent results.
  • No access — cannot query the item. It does not show up in Slite Agent results.
Workspace admins always have access, even when a root node is restricted.
In the root node list, the Access column reflects the current setting:
  • Everyone — anyone in the workspace can access it.
  • Limited — Everyone is set to No access, but specific users or groups have been granted access.
  • Admins — Everyone is set to No access and no exceptions have been added; only workspace admins can access it.

The Notion migration group

Notion sources previously followed mirrored access: a member only saw Notion results if their Slite email matched an account in the connected Notion workspace. Notion now uses Shared access, so admins choose who can query it from within Slite.
To make that switch safely, a user group was created automatically for each connected Notion workspace, named Notion migration ({Notion workspace name}) — so a Slite workspace with two Notion connections has two groups. Each contains the members whose email addresses were present in both Slite and that Notion workspace, and access was set to that group rather than to Everyone, so nobody gained access to Notion content they did not already have.
You can change this at any time: add or remove people from the group, replace it with specific users or groups of your choosing, or open the source up to everyone.
To change who can query Notion, set access explicitly on the connection rather than relying on the group alone. Removing the Notion migration group only removes that group's grant — it does not by itself grant access to anyone else.
This section only applies to Shared sources. Mirrored sources follow the original tool's permissions and can't be configured here.

Controlling who can add or manage sources

This is separate from the access roles above. It controls who can create and configure connections — the connection itself, not the content behind it. Someone can be allowed to add a Google Drive source without being able to query anything inside it.
This setting lives under Settings → Security → Create and manage sources.
Only workspace owners and admins can change this setting.

The Agent Sources panel

Open the Agent Sources panel from the left sidebar (at the bottom) to view all connections, add new ones, and manage what's syncing.
Every connection is listed individually — so Asana – Dev account and Asana – Marketing appear as separate entries rather than grouped. Click any connection to see its synced root nodes, access badges, and last sync times, or use + New source to add a new connection.

About sync timestamps

A "Last synced" timestamp like 17 hours ago doesn't mean the Slite Agent has been idle. It means nothing new changed in the source during that window, so there was nothing to update. The Slite Agent keeps checking in the background, and the timestamp only updates when new content is actually indexed.

Data Handling and Privacy

What gets indexed
The Slite Agent securely indexes the content needed for search and AI responses. What's indexed depends on each source's capabilities and permissions.
AI and privacy
Only the snippets relevant to a query are sent to the model. The model doesn't retain or train on your data.
Removing data
Disconnecting a source permanently deletes all its indexed content from the Slite Agent.
Security and compliance
All data is encrypted, stored in the EU, and handled to SOC 2 Type II standards. See the  Security Policy  and  Privacy Policy  for details.

Rate limits

Every tool the Slite Agent connects to has its own API rate limits. The Slite Agent manages these automatically — slowing or pausing sync when limits are hit, retrying once they reset, and resuming as soon as more requests are allowed. Previously indexed content stays searchable throughout.
Most of the time you won't notice. For very large sources or historical imports, you may see slower initial syncs or short delays in very recent content. No action required — the Slite Agent catches up automatically.

Removing a connection

To remove any connected source, click on it in the Agent Sources panel, then click the menu in the top right of the connection panel and select Remove connection. This permanently deletes all indexed content from that source.
For Files, remove individual files by clicking the three-dot menu next to the file name and selecting Remove source.
For Websites and Custom Sources, remove individual entries the same way — click the three-dot menu next to the source and select Remove source.