Atlassian With Mcporter

Atlassian With Mcporter

Use mcporter first for Atlassian work. Do not conclude that an issue/page is missing from a wrapper failure alone.

Workflow

1. Inspect the server before fallback:

   /Users/tao.exe/.nvm/versions/node/v22.22.2/bin/mcporter list atlassian --schema --json
   

2. If it fails with listen EPERM, rerun outside the sandbox with approval. The Atlassian MCP proxy may need local port binding.

3. Get accessible resources:

   /Users/tao.exe/.nvm/versions/node/v22.22.2/bin/mcporter call atlassian.getAccessibleAtlassianResources
   

4. For Jira issues, use the Atlassian MCP JQL tool:

   /Users/tao.exe/.nvm/versions/node/v22.22.2/bin/mcporter call atlassian.searchJiraIssuesUsingJql \
     cloudId=bankx.atlassian.net \
     jql='key = ISSUE-123' \
     maxResults=10 \
     fields='["summary","description","status","issuetype","priority","created","updated","project","assignee","reporter"]' \
     responseContentFormat=markdown
   

5. If search returns an ARI or lightweight result, call atlassian.fetch for details.

Guardrails

• Do not trust mcp__Confluence_wiki__.execute_jql_search alone; it may use obsolete Jira APIs.
• Do not say "issue not found" until Atlassian MCP JQL has been tried.
• If REST fallback is needed, prefer: https://api.atlassian.com/ex/jira/{cloudId}/rest/api/3/... over only: https://site.atlassian.net/rest/api/3/...
• Report failure causes separately: obsolete wrapper API, sandbox port failure, auth/permission failure, empty query result.