Confluence
ProductivityConnect AI agents to Confluence Cloud. Agents create and update pages and blog posts, manage comments, labels, and attachments, and search content with CQL across spaces.
What This Integration Enables
A wiki's failure mode is not emptiness, it is staleness: the page that says the old process is the current one. Confluence gives teams the writing surface; it does not keep the writing true. FlowRunner agents take over the mechanical half of that job. They create and update pages with versions incremented automatically, attach files, apply labels, announce changes in comments, and sweep the site with CQL to find what has gone quiet, so the humans only touch the wiki to say things, not to maintain it. - Publish and update pages and blog posts programmatically, versions handled for you - Sync documents into Confluence as attachments and pull them back out in later steps - Build knowledge-base structure: child pages under parents, labels for findability - Monitor and moderate discussion across footer and inline comments - Surface content across spaces with CQL or simple text, space, and type filters Deletion is the one move agents never make alone: FlowRunner's [human-in-the-loop](/concepts/human-in-the-loop) orchestration keeps a person between a stale page and the trash.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Every issue gets its page
When a new issue fires from [Jira](/integrations/jira-issues), the agent calls Create Page to publish a linked spec page under the project's parent, seeded with the issue summary and a structured template in storage format. Add Labels to Page tags it for the project, and Create Footer Comment notifies the assignee with what sections need their input. When the issue closes, Update Page stamps the outcome. The tracker and the wiki stop diverging because no human was supposed to keep them aligned.
The document pipeline into the knowledge base
A finished document is exported from [Google Docs](/integrations/google-docs). The agent calls Upload Attachment to place the rendered file on the relevant Confluence page, updates the page body with a summary and link, and posts a footer comment announcing the new version. Downstream flows can call Download Attachment to reuse the same file, so Confluence becomes the distribution point instead of the fifth place a copy lives.
The stale-content sweep with a named owner
Quarterly, the agent runs Search Content (CQL) for pages in the operations space untouched in a year, walks their trees with Get Child Pages, and assembles a review list with page owners resolved from version history. The list posts to [Slack](/integrations/slack). Owners mark keep, update, or remove. Only after that review does the agent act: Update Page for corrections, Delete Page to trash for removals. The wiki sheds dead weight on a schedule, and nothing disappears because an automation guessed.
Human-in-Loop Highlight
Delete Page has two modes, and the second is the one that matters: by default a deleted page moves to the trash and can be restored, but with Purge enabled it is permanently removed, and Delete Comment is permanent the moment it runs, taking every reply with it. A knowledge base is institutional memory; purging the wrong page can erase the only written record of why a decision was made. So FlowRunner agents run cleanup in two acts. Act one, reversible: the sweep trashes candidates and posts the list. Act two, permanent: only pages a documentation owner explicitly confirms get purged, and comment deletions in moderation flows carry the comment text in the approval request so the person sees exactly what disappears. The agent does the finding and the filing. A person owns everything that cannot come back.
Agent Capabilities
27 actionsSpaces
2- List Spaces Lists visible spaces with type and status filters, cursor-paginated up to 250 per page. The starting point for site-wide sweeps.
- Get Space Retrieves a space by ID with key, name, type, status, and homepage, optionally including the description. The placement step before page creation.
Pages
7- List Pages Lists pages site-wide with space, status, and exact-title filters, cursor-paginated. Bodies are not included; use Get Page for content.
- Get Page Retrieves a page with title, space, parent, version, and body in Storage, Atlas Doc Format, or View. The read behind reviews and updates.
- Get Pages in Space Lists all pages in a specific space with status filtering and sorting, cursor-paginated. The scoped sweep.
- Get Child Pages Lists a page's direct children, one level down. The tree-walking step for structured audits.
- Create Page Creates a page in a space, optionally under a parent, body in storage format (simple HTML works), as published or draft. The write behind issue-linked and templated pages.
- Update Page Updates title, body, or status, fetching the current page first and incrementing the version automatically. The maintenance write.
- Delete Page Deletes a page: to the trash by default (recoverable), permanently with Purge. The operation this page's human gate exists for.
Blog Posts
3- List Blog Posts Lists blog posts site-wide with space and status filters, cursor-paginated. Bodies via Get Blog Post.
- Get Blog Post Retrieves a blog post with title, space, version, and body in the chosen format.
- Create Blog Post Creates a blog post in a space, dated today when published as Current, body in storage format. Used for announcement and changelog automation.
Comments
5- List Footer Comments on Page Lists the regular comments below a page's content, with bodies, cursor-paginated. The read behind discussion monitoring.
- Create Footer Comment Adds a footer comment to a page, or a threaded reply when a parent comment ID is provided. The announcement step for page changes.
- Get Comment Retrieves a single footer comment with body, author, version, and parent.
- Delete Comment Permanently deletes a footer comment and its replies. Runs only behind explicit review in moderation flows.
- List Inline Comments on Page Lists comments anchored to highlighted text, with bodies and resolution status. Used to find unresolved review threads.
Attachments
5- List Page Attachments Lists a page's attached files with filename, media type, size, and download link, optionally filtered by exact filename.
- Get Attachment Retrieves an attachment's metadata by ID: filename, media type, size, owning page, and download link.
- Upload Attachment Uploads a FlowRunner file to a page as a minor edit, so watchers are not notified. Fails if the filename already exists on the page.
- Download Attachment Downloads an attachment into FlowRunner file storage and returns the saved file's URL for use in later steps.
- Delete Attachment Deletes an attachment: to the trash by default, permanently with Purge. Treated with the same two-act discipline as pages.
Labels
3- Get Page Labels Lists a page's labels, optionally filtered by prefix, cursor-paginated. The read behind taxonomy audits.
- Add Labels to Page Adds one or more global-prefix labels to a page, leaving existing labels intact. The findability step.
- Remove Label from Page Removes a single global-prefix label from a page; the label survives elsewhere.
Search and Identity
2- Search Content (CQL) Searches content with raw CQL or convenience text, space, and type filters, returning ranked matches with excerpts. The engine behind stale-content sweeps.
- Get Current User Retrieves the connected account's profile: account ID, display name, and email when visible. The attribution check.
Frequently Asked Questions
What can FlowRunner do with Confluence?
FlowRunner agents can run List Spaces, Get Space, and List Pages in Confluence, plus 24 more actions.
Does connecting Confluence to FlowRunner require OAuth?
Yes. Confluence connects to FlowRunner with OAuth 2.0, so agents authenticate without handling raw credentials.
Can Confluence trigger a FlowRunner workflow automatically?
Confluence doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with Confluence
$100 in credits. No card required. Connect in minutes.