Confluence integration. Manage document management data, records, and workflows. Use when the user wants to interact with Confluence data.
The confluence skill provides Confluence integration for managing document-management data, records, and workflows, letting an agent create, organize, and interact with Confluence content such as spaces, pages, attachments, and blog posts. Confluence itself is a team collaboration and knowledge-sharing hub for project documentation, meeting notes, and organizational knowledge, and this skill exposes that functionality to an agent so it can act on Confluence data when a user asks.
The skill works through the Membrane CLI, which handles authentication and automatic credential refresh so users never supply API keys directly. Setup involves installing the CLI with npm install -g @membranehq/cli@latest, authenticating with membrane login --tenant --clientName=<agentType> (completing membrane login complete <code> in headless environments), and establishing a connection with membrane connection ensure, polling connection state (BUILDING, READY, CLIENT_ACTION_REQUIRED, CONFIGURATION_ERROR, SETUP_FAILED) via membrane connection get <id> --wait as needed. Agents discover functionality with membrane action list --connectionId=CONNECTION_ID --intent "QUERY" and run actions with membrane action run <actionId> --connectionId=CONNECTION_ID, passing parameters via --input. Popular actions include list-pages, list-blog-posts, list-spaces, get-page, create-page, create-blog-post, create-space, create-page-comment, update-page, update-task, and delete-page. When pre-built actions do not cover a use case, membrane request CONNECTION_ID /path/to/endpoint proxies calls directly to the Confluence API with automatic base-URL and auth-header injection, supporting flags like -X for method, -H for headers, -d/--json for body, and --query and --pathParam for parameters. Adding --json to any command yields machine-readable output.
The skill targets teams and agents that need to automate Confluence document management. Its stated compatibility requirement is network access and a valid Membrane account, with the Free tier supported, and it is MIT licensed. Best practices include preferring Membrane over raw API calls for built-in auth, pagination, and error handling, discovering existing actions before writing custom calls, and letting Membrane manage credentials server-side rather than asking users for tokens.
It requires network access and a valid Membrane account (the Free tier is supported). You install the Membrane CLI with npm install -g @membranehq/cli@latest and authenticate via membrane login; Membrane then handles authentication and credential refresh automatically.
Run membrane connection ensure "https://www.atlassian.com/software/confluence" --json, which finds or creates a connection and returns a connection id. If the connection is BUILDING, poll with membrane connection get <id> --wait --json until the state becomes READY (or handle CLIENT_ACTION_REQUIRED, CONFIGURATION_ERROR, or SETUP_FAILED).
Search with membrane action list --connectionId=CONNECTION_ID --intent "QUERY" to discover actions by natural-language intent, then run one with membrane action run <actionId> --connectionId=CONNECTION_ID, passing parameters via --input '{"key": "value"}'. The result is in the output field of the response.
Use the proxy: membrane request CONNECTION_ID /path/to/endpoint sends requests directly to the Confluence API, with Membrane appending the base URL and injecting auth headers (including credential refresh). Options include -X for method, -H for headers, -d or --json for the body, and --query and --pathParam for parameters.
No. A best practice of the skill is to never ask the user for API keys or tokens; you create a connection and Membrane manages the full auth lifecycle server-side with no local secrets.
Quick Setup:
.claude/skills/Repository
membranedev/application-skills