Confluence Data Source

Connect Confluence pages to Agent Factory so RAG Agents can retrieve approved product documentation, procedures, and other organizational content.

A Confluence data source crawls a configured root page and its descendants. Agent Factory supports Confluence Cloud and Confluence Data Center through OAuth authorization-code authentication.

Prepare Confluence OAuth Access

Before you create the data source:

1. Create an OAuth application for the target Confluence deployment.

2. Request and authorize the following scopes:

   • offline_access: Obtain a refresh token for scheduled crawls without another interactive authorization.
   • read:page:confluence: Read page content and metadata.
   • read:hierarchical-content:confluence: Enumerate descendants of the configured root page.
   • read:space:confluence: Resolve the page’s space key and metadata.
3. Complete the OAuth authorization-code flow with an account that can read the intended site, root page, and descendant pages.
4. Record the client ID, client secret, and refresh token for the data-source configuration.

Grant the OAuth account access only to the Confluence content that Agent Factory needs to ingest.

Configure a Confluence Data Source

1. Create a data source and select Confluence as the source type.
2. Enter a source name and description.
3. Select Cloud or Data Center as the deployment type.
4. Enter the URL of the root Confluence page to crawl.
5. Enter the OAuth client ID, client secret, and refresh token.

6. Optionally, configure the content filters:

   • Include filters ingest content whose name contains at least one configured substring.
   • Exclude filters omit content whose name contains a configured substring.
   • Exclude file extensions remove the specified extensions from the supported file-extension set.
7. Set the crawl depth to limit how many descendant-page levels Agent Factory follows from the root page.
8. Set the crawl frequency in days.
9. Optionally, provide a proxy URL for Confluence and OAuth requests. See Configure Proxy.
10. Create the data source and monitor its ingestion status.

When the status is ingested, you can select the Confluence data source while creating or editing a RAG Agent. See Create RAG Agent.

Agent Factory protects the OAuth client secret in Oracle Wallet and stores the refresh token in encrypted form. Scheduled crawls use the refresh token to obtain new access tokens.

Control Ingested Content and Access

The crawler ingests Confluence pages only. It does not ingest file attachments.

Warning: Confluence page permissions are not carried into Agent Factory. After ingestion, RAG Agent access controls govern who can retrieve the content. A person with access to the RAG Agent might retrieve content without having direct access to the source page in Confluence.

Before ingestion:

• Select a root page whose descendants are appropriate for all intended RAG Agent users.
• Limit the OAuth account’s access to the pages that Agent Factory should crawl.
• Review page content for privacy and data-governance requirements.
• Grant RAG Agent access only to authorized users and groups. See Manage Identity and Access.
• Use only a trusted proxy because authentication and page requests can pass through it.

Troubleshoot Confluence Sources

Symptom	Check
OAuth authentication fails	Verify the client ID, client secret, refresh token, authorized scopes, and OAuth application configuration.
Confluence Cloud site discovery fails	Confirm that the root-page hostname identifies one site that the OAuth account can access.
Expected pages are missing	Verify root-page and descendant access, crawl depth, and include and exclude filters.
File attachments are missing	File attachments are not supported for Confluence data-source ingestion.
Proxy-related failures occur	Verify the proxy URL and the global proxy settings in Configure Proxy.

See Data Sources Troubleshooting for general ingestion statuses and errors.