Confluence Data Center in Claude, ChatGPT and any MCP client. For your whole team.
CQL search, pages, comments, attachments, spaces. One token in Loopthink, one login per person, roles per tool and an audit entry for every call.
Free for 2 connectors, no card · Pro €19 · EU-hosted · 13 read tools, 2 write tools off by default
“Which pages in the PROJ space were changed this week?”
Three pages in PROJ were changed this week:
| Page | Parent | Version | Updated |
|---|---|---|---|
| Rollout Kraus & Partner branch | Projects 2026 | 14 | 23 Oct 2026 |
| Steering meeting 21 Oct | Meetings 2026 | 3 | 21 Oct 2026 |
| Interface Nordwind Handel | Integrations | 8 | 20 Oct 2026 |
- read-only
- masked
One token in, the tools come out.
Confluence Data Center is the intranet and documentation of many larger companies in the DACH region, on their own servers: process descriptions, project pages, meeting notes, how-tos, with comments, attachments and labels. The knowledge is there; finding the current page among five similar ones is the hard part.
AI connectors for Confluence mostly cover the cloud edition, so teams on Data Center have been left out. With the Confluence Data Center connector, Operations, HR, Engineering and Project management ask in Claude or ChatGPT and get the answer from their own pages. Loopthink holds one personal access token of a Confluence user, every person signs in with their own identity, and every call is logged under their name.
Thirteen read tools cover CQL search, pages and blog posts with their body, page trees, history, comments, attachments, labels, spaces and users. Two write tools, a comment and a new page, stay off until you switch them on for a role.
- Operations
“How does our onboarding process for new suppliers work? Find the current page.”
Answered by
search_content - HR
“Summarise the travel expense policy for me, with the limits per night.”
Answered by
get_content - Project management
“Which pages in the PROJ space were changed this week, and by whom?”
Answered by
search_content - Engineering
“Show me the page tree below Architecture in the IT space.”
Answered by
list_page_children - Operations
“Who last edited the emergency plan, and when?”
Answered by
get_content_history - Project management
“Create a page with the minutes of today's steering meeting below Meetings 2026.”
Answered by
create_page
What the Confluence Data Center connector can do.
| Tool | What it does | Access | Default |
|---|---|---|---|
| get_current_user | Returns the user behind the token (username, userKey, displayName). Use to test the connection. | Read | On |
| search_content | Searches pages, blog posts, comments and attachments with CQL, the main way to find content. Examples: 'type = page AND text ~ "onboarding"', 'space = DEV AND title ~ "release n… | Read | On |
| get_content | Returns one page, blog post or comment by id with its body. Default expand gives the body in storage format (XHTML with Confluence macros), the version, the space and the ancest… | Read | On |
| find_content_by_title | Finds pages by exact title in a space, or lists pages or blog posts of a space. For a page title pass space_key and title; for blog posts pass type=blogpost and optionally posti… | Read | On |
| list_page_children | Lists the direct child pages of a page (id, title). Use to walk a page tree; call again on a child to go deeper. | Read | On |
| get_content_history | Returns the history of a page or blog post: who created it and when, the last update (by, when, version number, message) and the previous version. Use to answer 'who changed thi… | Read | On |
| list_comments | Lists the comments of a page or blog post with author and body. Footer and inline comments are both returned unless location narrows it; depth=all includes replies. | Read | On |
| list_attachments | Lists the attachments of a page or blog post: id, title (file name), media type, file size and the download link in _links.download (relative to the instance URL). Paged with st… | Read | On |
| list_labels | Lists the labels of a page or blog post (prefix and name). Use label names in CQL like label = howto. | Read | On |
| list_spaces | Lists the spaces the user can view (id, key, name, type global or personal). Call first to get a space key for CQL. | Read | On |
| get_space | Returns one space by key with its description and homepage (the root page id, a good starting point for list_page_children). | Read | On |
| list_space_content | Lists the pages or blog posts of a space. depth=root returns only the top-level pages, all (default) every page. | Read | On |
| get_user | Returns a user by username or user key (displayName, username, userKey). Use to resolve the userKey or username found in a page history or comment author. | Read | On |
| add_comment | WRITE: Adds a footer comment to a page or blog post as the token's user. The body is Confluence storage format (XHTML), e.g. | Write | Off |
| create_page | WRITE: Creates a new page in a space below a parent page. For a top-level page use the space homepage id from get_space as parent. | Write | Off |
Write tools are off for every role until you switch them on, one tool for one role. The names are what the model sees; it also gets the full parameter description of each tool.
Getting the Confluence Data Center token.
Five minutes, once. Nobody on your team ever sees it.
- 1
Log in to Confluence as the user the AI should act as (ideally a dedicated technical user with view permission on exactly the spaces the AI may read).
- 2
Click your avatar (top right) > Settings, then Personal Access Tokens.
- 3
Click Create token, enter a name and an expiry in days (your admin may enforce a maximum, 365 days by default), then Create.
- 4
Copy the token immediately. It is not shown again after you close the dialog.
- 5
In Loopthink, add Confluence Data Center from the catalog and paste the token. It also asks for confluence url. Your team sees the tools a minute later.
Personal access tokens exist from Confluence 7.9 and have no scopes: a token acts with all permissions of its user, including restricted pages that user can see. Give a technical user only the space permissions the AI needs; add_comment and create_page additionally need the Add Comments or Add Pages space permission. Admins can disable tokens or cap their lifetime with the atlassian.pats.* system properties.
What we suggest masking in Confluence Data Center.
Pages are free text: meeting notes name people, HR pages carry personal details, and every page shows who wrote and changed it. The suggested fields mask display names, usernames, e-mail addresses and authors, and for roles that should only find documents also titles, excerpts and bodies. Better still, give the token a user that cannot see the HR spaces at all.
- displayName
- username
- userKey
- by
- createdBy
- creator
- author
- contributors
- profilePicture
- body
- value
- excerpt
- title
Confluence Data Center and Loopthink
- Which Confluence user and permissions does the token need?
- A personal access token, available from Confluence 7.9. It has no scopes and acts with all permissions of its user, including restricted pages that user can see. Create it under a technical user with view permission on exactly the spaces the AI may read. add_comment and create_page additionally need the Add Comments or Add Pages space permission.
- Atlassian has announced the end of Data Center. Is this still worth it?
- Atlassian has announced the end of Data Center support for March 2029. Until then, your instance holds the knowledge your teams look up every day, and the connector makes it usable in Claude and ChatGPT now, without a migration. If you move to Cloud later, the roles, masking rules and Audit-Log stay in Loopthink.
- Our Confluence runs in our own data center. Does the connector reach it?
- The instance has to be reachable from the internet over HTTPS, for example through your reverse proxy. If it is only reachable in your network, tell us and we look at it with you.
- Does it read attachments?
- It lists them with file name, type, size and download link. The content of attached files is not returned; page and blog post bodies are.
- Does every team member need their own Confluence Data Center account?
- No. Loopthink holds one key or token for Confluence Data Center. Your people sign in to Loopthink with their own identity, and every call to Confluence Data Center is logged under their name. Seats in Confluence Data Center stay where they are today.
- Where does the Confluence Data Center key live?
- In Loopthink, encrypted, in the EU. It never reaches the AI client and never reaches the people using the connector. Rotate it in Confluence Data Center whenever you like and paste the new one once.
- Can somebody write to Confluence Data Center from the chat?
- Not unless you switch it on. Write actions are a separate row in the permission matrix and are off for every role by default. If a role needs one write tool, you grant that one tool to that one role.
- What does the model see?
- The result of the tool call, after masking. Fields you mark as personal, such as emails, names or card details, are masked before the answer travels to the AI client. The model gets the numbers, not the people.
- What does it cost?
- Free for one person and two connectors, no card. Pro is €19 a month for up to five people and every connector on the site, with roles, field masking and a 30-day audit. Team is €49 for ten people and adds your own REST APIs and SQL databases, your own domain and your own identity provider. No call needed.
Stuck, or a tool you need is missing? A person who knows this connector answers within one working day.
