Skill by ara.so — MCP Skills collection.

• Search & Summarize: Query Jira issues, Confluence pages, and Compass components
• Create & Update: Generate issues, pages, and components from natural language
• Automate Workflows: Link content, bulk create items, extract data across Atlassian products
• Respect Permissions: All actions honor existing Atlassian Cloud access controls

1. Open a browser to Atlassian OAuth consent screen
2. Prompt you to authorize access to Jira/Confluence/Compass
3. Store credentials securely for future sessions

{
  "mcpServers": {
    "atlassian-rovo": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/mcp-remote",
        "https://mcp.atlassian.com/v1/mcp"
      ],
      "env": {
        "MCP_REMOTE_AUTH_TYPE": "apitoken",
        "ATLASSIAN_EMAIL": "${ATLASSIAN_EMAIL}",
        "ATLASSIAN_API_TOKEN": "${ATLASSIAN_API_TOKEN}",
        "ATLASSIAN_CLOUD_ID": "${ATLASSIAN_CLOUD_ID}"
      }
    }
  }
}

• ATLASSIAN_EMAIL

  : Your Atlassian account email

• ATLASSIAN_API_TOKEN

  : Rovo MCP scoped API token (from Atlassian account settings)

• ATLASSIAN_CLOUD_ID

  : Your site URL (e.g.,

  https://yoursite.atlassian.net

  )

• Server URL:

  https://mcp.atlassian.com/v1/mcp

• Auth Type: OAuth 2.1 (browser flow)

1. Go to Settings → Integrations → MCP
2. Add server URL:

   https://mcp.atlassian.com/v1/mcp

3. Complete OAuth flow when prompted

• MUST use Jira project key = MYPROJ
• MUST use Confluence spaceId = "123456"
• MUST use cloudId = "https://mysite.atlassian.net" (do NOT call getAccessibleAtlassianResources)
• MUST use

  maxResults: 10

  or

  limit: 10

  for ALL Jira JQL and Confluence CQL search operations.

Replace with your actual values:
- `MYPROJ`: Your default Jira project key
- `123456`: Your Confluence space ID (get from space settings)
- `https://mysite.atlassian.net`: Your Atlassian Cloud site URL

"Read the 'Feature Requests' Confluence page and create a Jira ticket for each item"

1. Fetches Confluence page content
2. Parses items
3. Creates Jira issues using

   jira_createIssue

• read:jira-work

  : Read Jira data

• write:jira-work

  : Create/update Jira issues

• read:confluence-content.all

  : Read Confluence pages

• write:confluence-content

  : Create/update Confluence pages

• read:compass:*

  : Read Compass data

• write:compass:*

  : Write Compass data

1. Ensure your current IP is allowed in Atlassian Administration
2. Requests through MCP server must originate from allowed IPs
3. Configure VPN if working remotely

• jira_searchForIssuesUsingJql

• jira_createIssue

• jira_editIssue

• jira_addComment

• jira_getIssue

• jira_deleteIssue

• jira_assignIssue

• confluence_searchByCQL

• confluence_createPage

• confluence_updatePage

• confluence_getPage

• confluence_deletePage

• confluence_getSpaces

• compass_createComponent

• compass_searchComponents

• compass_getComponent

• compass_updateComponent

• compass_deleteComponent

• getAccessibleAtlassianResources

  (list available sites - avoid if using AGENTS.md defaults)

1. Have a site admin complete OAuth flow first
2. Once installed, regular users can connect

1. Check Atlassian Administration → Security → IP allowlist
2. Add your IP range or VPN IP
3. Contact your admin if you can't modify settings

1. Switch to API token authentication (see headless setup above)
2. Or ensure

   BROWSER

   environment variable points to valid browser

1. Verify you're logged into correct Atlassian account
2. Check site URL matches

   cloudId

   in config
3. Ensure you have access to Jira/Confluence/Compass on that site
4. Try revoking and re-authorizing from Atlassian account settings

getAccessibleAtlassianResources

1. Add AGENTS.md configuration with

   cloudId

   , project, and space defaults
2. Always specify

   maxResults

   or

   limit

   in search queries
3. Cache site/project/space IDs in conversation context

1. Batch operations where possible
2. Add delays between bulk creates
3. Use search with pagination instead of fetching all results

skills/

• create-jira-issue.md: Create Jira issues from natural language
• search-confluence.md: Search and summarize Confluence pages
• link-content.md: Link Jira issues to Confluence pages

1. Copy skill file to

   ~/Library/Application Support/Claude/skills/

2. Restart Claude Desktop
3. Reference skill: "Use the create-jira-issue skill to make a new bug"

1. First user must have access to all Atlassian products being integrated (Jira, Confluence, Compass)
2. Complete OAuth flow to install app (lazy/JIT installation)
3. App appears in Atlassian Administration → Apps → Connected apps

1. Go to Atlassian Administration → Security → Rovo MCP Server
2. Enable "API token authentication"
3. Users can then create scoped API tokens from account settings

• View audit logs in Atlassian Administration → Audit log
• Filter by "Rovo MCP Server" to see all actions
• Logs include user, timestamp, action, IP address

1. Atlassian Administration → Apps → Connected apps
2. Find "Atlassian Rovo MCP Server"
3. Click "Revoke" or "Uninstall"

1. User profile → Account settings → Security
2. Connected apps → Revoke "Atlassian Rovo MCP Server"

1. Never commit API tokens: Use environment variables (

   ${ATLASSIAN_API_TOKEN}

   )
2. Use scoped tokens: Request only needed permissions
3. Rotate tokens regularly: Refresh API tokens every 90 days
4. Monitor audit logs: Review MCP actions weekly
5. Enable IP allowlisting: Restrict access to known networks
6. Educate users: Train on what data agents can access
7. Revoke unused access: Remove old OAuth authorizations

• Official Documentation
• Admin Controls
• Monitoring Guide
• API Token Setup
• Supported Domains