confluence

name	confluence
description	Manage Confluence pages with search, read, and write operations
allowed-tools	Bash, Read

Confluence Skill

Rovo-first guide for Confluence pages, with the local Sidekick client kept as a fallback.

Try Rovo first for reads and writes. The local Sidekick Confluence client is for fallback, debugging, local cache behavior, or raw-storage compatibility work.

When invoked, use Atlassian Rovo MCP first to handle the request: $ARGUMENTS

Primary Path: Atlassian Rovo MCP

• Natural-language discovery: use Rovo search.
• Page details from search results: fetch by returned Atlassian resource ID when available.
• Direct page URL or page ID: use Rovo page fetch when Markdown or ADF is sufficient.
• Page writes: use Rovo create/update page tools, after the normal remote-write confirmation rule is satisfied.

Confluence Update Precedence

For Confluence page updates, use this fallback order:

1. Atlassian Rovo MCP, when the required complete page body can be fetched safely.
2. Chrome plugin/live-editor automation, especially for large pages or cases where Rovo cannot safely read the full body.
3. Local Sidekick Confluence client / Confluence REST API raw-storage HTML, only when ATLASSIAN_API_TOKEN is set.

Large Page Limitation

Do not use Rovo for read-modify-write edits on large Confluence pages unless you have verified the Rovo page body is complete. Rovo page updates replace the full page body, while Rovo page reads can truncate large bodies and Rovo does not currently expose partial page updates or paginated full-page reads. A full-body write based on a truncated read can delete the unread tail of the page.

For large-page edits, prefer Chrome plugin/live-editor automation after Rovo. Use the local Sidekick Confluence client with raw storage HTML from the Confluence REST API only when ATLASSIAN_API_TOKEN is set. The token may rotate every 3 days.

Use the local Sidekick client when Rovo and Chrome plugin paths are unavailable or unsuitable, the needed API token is set, raw Confluence storage HTML is required, local cache behavior is specifically useful, debugging the client itself, or the user explicitly asks for the local client.

Fallback note: Local Confluence commands return Markdown by default. Use --html only when raw storage HTML is required for content manipulation.

Sidekick CLI Fallback Commands

Search for Pages

python3 -m sidekick.clients.confluence search "query" [--space SPACE] [--limit N]

Get Page Details

python3 -m sidekick.clients.confluence get-page PAGE_ID_OR_URL
python3 -m sidekick.clients.confluence get-page-from-link "CONFLUENCE_URL"

Get Page by Title

python3 -m sidekick.clients.confluence get-page-by-title "Title" SPACE

Read Page Content

# Fallback for Confluence URLs when Rovo cannot be used.
# Supports full page URLs and tiny URLs:
# - https://domain.atlassian.net/wiki/spaces/SPACE/pages/123456/Title
# - https://domain.atlassian.net/wiki/x/SHORTID
python3 -m sidekick.clients.confluence get-content-from-link "CONFLUENCE_URL"

# For raw HTML from a URL (content manipulation)
python3 -m sidekick.clients.confluence get-content-from-link "CONFLUENCE_URL" --html

# Read by page ID or URL; returns Markdown by default
python3 -m sidekick.clients.confluence read-page PAGE_ID_OR_URL

# For raw HTML (content manipulation)
python3 -m sidekick.clients.confluence read-page PAGE_ID_OR_URL --html

Create New Page

python3 -m sidekick.clients.confluence create-page SPACE "Title" content.html [--parent PARENT_ID]

Update Page

python3 -m sidekick.clients.confluence update-page PAGE_ID content.html [--title "New Title"]

For meeting-note agenda updates, use the confluence-meeting-notes-update skill. It performs safe Rovo ADF edits with structural validation. For creating or preparing the next meeting notes section, use the confluence-meeting-notes-create-next skill.

Search Cache

The local Confluence client automatically caches search query to page mappings for faster repeated searches.

Cache Management:

python3 -m sidekick.clients.confluence cache-show
python3 -m sidekick.clients.confluence cache-clear

Example Usage

When the user asks to:

• "Search for API documentation in Confluence" - Use Rovo search first; fall back to the search command only if needed
• "Read the contents of my 1:1 doc with Bob" - Use Rovo search/fetch first; fall back to the local client only if needed
• "Read this Confluence URL" - Use Rovo page fetch first; fall back to get-content-from-link for full page URLs and /wiki/x tiny URLs
• "Add a topic to my 1:1 with Alice" - Use the confluence-meeting-notes-update skill
• "Create the next section for this meeting doc" - Use the confluence-meeting-notes-create-next skill
• "Update the team wiki page" - Use Rovo update page first; fall back to update-page when raw storage HTML is required or Rovo cannot be used

For full documentation, see the detailed Confluence skill documentation in this folder.