Jira Confluence

Pre-flight Check (Execute First)

Before any Jira/Confluence operation, verify the MCP server is available.

Hard Rules

• If the user asks for Jira or Confluence content, always use Atlassian MCP tools first.
• Do not use curl, generic web access, browser scraping, or raw HTML fetching before Atlassian MCP has been attempted.
• If the user provides a Confluence URL, prefer resolving it into page_id or space_key + title and call confluence_get_page(...) directly.
• If Atlassian MCP is unavailable, stop the content-retrieval flow and enter setup guidance. Do not substitute web scraping for MCP-backed reads.
• For Jira/Confluence requests, this skill takes priority over generic web-fetch behavior.

Step 1: Check MCP Tools

Check whether Atlassian MCP tools are available in the current session.

• Tools found (for example mcp__mcp_atlassian__confluence_get_page, mcp__mcp_atlassian__confluence_search, mcp__mcp_atlassian__jira_search) → Proceed to operations
• No tools found → Go to Step 2

Step 2: Setup Required

Tell the user:

"The Jira/Confluence MCP server is not configured. Would you like me to help you set it up? It takes about 2 minutes."

If user agrees, read references/setup.md and guide them through:

1. Install uv/uvx (if needed)
2. Configure MCP settings with Agora OAuth credentials
3. Restart AI tool - browser will open for OAuth SSO login
4. Verify connection

Step 3: Verify Connection

After setup, test with a simple query:

jira_search(jql="assignee = currentUser() AND statusCategory != Done ORDER BY updated DESC", limit=1)

• Success → Ready to use
• Error → Check references/setup.md troubleshooting section

Jira Operations

Search Issues

Use jira_search with JQL. Common patterns:

# My open issues (use statusCategory for i18n support)
assignee = currentUser() AND statusCategory != Done ORDER BY updated DESC

# Project backlog
project = PROJ AND statusCategory = "To Do" ORDER BY priority DESC

# Recent bugs
project = PROJ AND type = Bug AND created >= -7d

# Issues in current sprint
sprint in openSprints() AND assignee = currentUser()

# High priority unresolved
assignee = currentUser() AND statusCategory != Done AND priority in (High, Highest)

Tips:

• Use statusCategory instead of status for better i18n support (works with Chinese/localized status names)
• Quote multi-word values: status = "In Progress"
• Use currentUser() for the logged-in user
• Add ORDER BY for sorting: ORDER BY updated DESC
• See references/jql-cql-reference.md for more advanced JQL/CQL patterns

Get Issue Details

Use jira_get_issue(issue_key="PROJ-123") to fetch full details including description, comments, and status.

Create Issues

Use jira_create_issue with required fields:

• project_key: Project key (e.g., "PROJ") - always ask user if unknown
• summary: Issue title
• issue_type: Type name (Task, Bug, Story, Epic, etc.)

Optional fields:

• description: Markdown content
• assignee: Email or display name
• components: Comma-separated names
• additional_fields: JSON for priority, labels, parent, epic link

Creating subtasks: (see references/jira-create-subtasks.md for details)

jira_create_issue(
    project_key="PROJ",
    summary="Subtask title",
    issue_type="Subtask",  # or "子任务" in Chinese instances
    additional_fields='{"parent": "PROJ-123"}'
)

Update Issues

Use jira_update_issue(issue_key="PROJ-123", ...) to modify:

• summary, description, assignee, priority, labels, etc.

Transition Status

1. Get available transitions: jira_get_transitions(issue_key="PROJ-123")
2. Execute transition: jira_transition_issue(issue_key="PROJ-123", transition="Done")

Add Comments

jira_add_comment(issue_key="PROJ-123", body="Comment in **Markdown**")

Confluence Operations

If User Provides a Confluence URL

Handle Confluence URLs directly through MCP, not through web fetching.

Preferred order:

1. If the URL contains pageId=..., call confluence_get_page(page_id="...")
2. If the URL is in /display/SPACE/TITLE form, extract space_key and title, then call confluence_get_page(title="...", space_key="...")
3. If title parsing is ambiguous, use confluence_search first, then fetch the matching page

Examples:

• https://confluence.agoralab.co/pages/viewpage.action?pageId=123456789
  • confluence_get_page(page_id="123456789")
• https://confluence.agoralab.co/display/DPE/Demo+2.2.0+HLD
  • confluence_get_page(title="Demo 2.2.0 HLD", space_key="DPE")

Search Pages

Use confluence_search with text or CQL:

# Simple text search
confluence_search(query="project documentation")

# CQL: pages in specific space
confluence_search(query="type=page AND space=TEAM")

# CQL: recently modified
confluence_search(query="lastModified > startOfWeek()")

Get Page Content

# By page ID
confluence_get_page(page_id="123456789")

# By title and space
confluence_get_page(title="Meeting Notes", space_key="TEAM")

Handling Jira Macros in Pages

Agora Confluence pages often embed Jira issue links via macros. These cause two problems:

Reading: Markdown conversion produces garbage like JIRA2cd8ad8e-4449-3a5a-9d76-fd5a49868868EEP-1484. The actual issue key is the suffix after the UUID — in this case EEP-1484. When presenting to the user, extract and show only the issue key.

Editing: Pages with Jira macros must be edited in storage format to preserve the macros. Using markdown format will destroy all Jira links and user mentions.

To edit a page with macros:

1. Read the page in raw HTML: confluence_get_page(page_id="...", convert_to_markdown=false)
2. Modify only the parts you need in the storage XML
3. Write back using storage format: confluence_update_page(..., content_format="storage")

To insert a new Jira issue link, use this storage XML template:

<ac:structured-macro ac:name="jira" ac:schema-version="1">
  <ac:parameter ac:name="server">JIRA</ac:parameter>
  <ac:parameter ac:name="serverId">2cd8ad8e-4449-3a5a-9d76-fd5a49868868</ac:parameter>
  <ac:parameter ac:name="key">PROJ-123</ac:parameter>
</ac:structured-macro>

The serverId 2cd8ad8e-4449-3a5a-9d76-fd5a49868868 is the Agora Jira application link ID — use it as-is for all Jira macros.

Create Pages

confluence_create_page(
    space_key="TEAM",
    title="New Page Title",
    content="# Heading\n\nMarkdown content here",
    parent_id="123456"  # optional
)

Update Pages

# Simple pages (no macros) — use markdown
confluence_update_page(
    page_id="123456789",
    title="Updated Title",
    content="# New Content\n\nUpdated markdown"
)

# Pages with Jira macros — use storage format
confluence_update_page(
    page_id="123456789",
    title="Updated Title",
    content="<p>Updated content with <ac:structured-macro ...>...</ac:structured-macro></p>",
    content_format="storage"
)

Quick Reference

Common Issues