Atlassian ITSM Skill

IT Service Management workflows using Jira for incident/change tracking and Confluence for runbooks, postmortems, and documentation.

Purpose

This skill provides ITSM-focused workflows for network operations teams using Atlassian tools:

• Jira: Incident tracking, change requests, problem management, and task coordination
• Confluence: Runbook documentation, postmortem reports, and knowledge base maintenance

All write operations require human-in-the-loop confirmation per Constitution XIV.

Jira Read Workflows (User Story 1)

Discovering Projects and Issue Types

Before creating or searching issues, discover the available projects and their issue types.

Step-by-step workflow:

1. List accessible projects: Call jira_get_projects to see all projects the authenticated user can access
2. Get project details: Call jira_get_project(project_key) for specific project configuration
3. List issue types: Call jira_get_issue_types(project_key) to see available issue types (Bug, Task, Story, etc.)
4. List available fields: Call jira_get_fields to see standard and custom fields

Example prompts:

• "Show me all Jira projects I have access to"
• "What issue types are available in project NET?"
• "List the custom fields in our Jira instance"

Searching Issues with JQL

Search issues using Jira Query Language (JQL) for powerful filtering.

Step-by-step workflow:

1. Construct JQL query: Build query using project, status, assignee, label, and date filters
2. Execute search: Call jira_search(jql_query, max_results) with the JQL string
3. Review results: Parse the returned issue list with key, summary, status, assignee

Common JQL patterns:

# Open issues in project NET assigned to me
project = NET AND assignee = currentUser() AND status != Done

# Critical/blocker issues created this week
project = NET AND priority IN (Critical, Blocker) AND created >= -7d

# Issues with specific label
project = NET AND labels = "outage"

# Unassigned issues in backlog
project = NET AND assignee IS EMPTY AND status = "To Do"

# Issues updated in the last 24 hours
project = NET AND updated >= -1d

Example prompts:

• "Show me open issues in project NET assigned to me"
• "Search for critical issues created this week in project OPS"
• "Find all issues with label 'network-outage'"

Retrieving Issue Details

Get full details for a specific issue by key.

Step-by-step workflow:

1. Get issue details: Call jira_get_issue(issue_key) with the issue key (e.g., "NET-1234")
2. Review issue data: Parse fields including summary, description, status, priority, assignee, reporter, labels, components, fix versions
3. Get comments: Call jira_get_issue_comments(issue_key) to see discussion history

Example prompts:

• "Get details for issue NET-1234"
• "Show me the comments on issue OPS-567"
• "What is the current status and priority of NET-1234?"

Confluence Read Workflows (User Story 2)

Discovering Spaces

Before searching or creating pages, discover available Confluence spaces.

Step-by-step workflow:

1. List all spaces: Call confluence_get_spaces to see accessible spaces
2. Get space details: Call confluence_get_space(space_key) for specific space info

Example prompts:

• "Show me all Confluence spaces I have access to"
• "Get details for the NET-DOCS space"

Searching Pages with CQL

Search Confluence pages using Confluence Query Language (CQL).

Step-by-step workflow:

1. Construct CQL query: Build query using space, title, text, and date filters
2. Execute search: Call confluence_search(cql_query, max_results)
3. Review results: Parse returned page list with title, space, ID

Common CQL patterns:

# Pages in space NET-DOCS containing "BGP"
space = "NET-DOCS" AND text ~ "BGP"

# Pages with title containing "runbook"
title ~ "runbook"

# Recently modified pages in a space
space = "NET-DOCS" AND lastModified > now("-7d")

Example prompts:

• "Search Confluence for 'BGP runbook'"
• "Find pages in NET-DOCS space containing 'OSPF'"
• "Show me recently modified pages in the network documentation space"

Retrieving Page Content

Get full page content by ID or title+space combination.

Step-by-step workflow:

1. Get page content: Call confluence_get_page(page_id) or confluence_get_page(space_key, title)
2. Review content: Parse body, version number, and metadata
3. Get comments: Call confluence_get_page_comments(page_id) to see page comments

Example prompts:

• "Show me the 'Router Maintenance Procedures' page in NET-DOCS"
• "Get the content of Confluence page ID 123456"
• "What comments are on the BGP runbook page?"

Jira Write Workflows (User Story 3)

Creating Issues

Create new Jira issues with human-in-the-loop confirmation.

Step-by-step workflow:

1. Query available types: Call jira_get_issue_types(project_key) to verify available issue types
2. Get available fields: Call jira_get_fields to understand required/optional fields
3. Present to operator: Display the proposed issue details and request confirmation (Constitution XIV)
4. Create issue: After confirmation, call jira_create_issue with project, issue_type, summary, description, priority, labels, assignee
5. Confirm creation: Return the created issue key to the operator

Example prompts:

• "Create a Jira issue in project NET titled 'Router-01 BGP peer down' with priority Critical"
• "Log an incident in project OPS for the core switch fan failure"

IMPORTANT: Always present the full issue details to the operator and wait for explicit confirmation before calling jira_create_issue.

Updating Issues

Update existing issue fields with human-in-the-loop confirmation.

Step-by-step workflow:

1. Get current state: Call jira_get_issue(issue_key) to see current field values (read-before-write per Constitution II)
2. Present changes: Display current vs proposed field values to operator
3. Update issue: After confirmation, call jira_update_issue(issue_key, fields) with the changed fields
4. Confirm update: Verify the update succeeded

Updateable fields: summary, description, priority, labels, assignee, components, fix_versions

Example prompts:

• "Update NET-1234 priority to Critical"
• "Add label 'outage' to issue OPS-567"
• "Assign NET-1234 to john.smith"

Transitioning Issues

Move issues through workflow states.

Step-by-step workflow:

1. Get available transitions: Call jira_get_transitions(issue_key) to see valid transitions from current state
2. Present options: Show operator the available transitions (e.g., "Start Progress", "Resolve", "Close")
3. Execute transition: After confirmation, call jira_transition_issue(issue_key, transition_id)
4. Confirm transition: Verify the new status

Example prompts:

• "Transition NET-1234 to In Progress"
• "Mark issue OPS-567 as Resolved"
• "What workflow transitions are available for NET-1234?"

Adding Comments

Add comments to issues for investigation updates.

Step-by-step workflow:

1. Draft comment: Prepare comment text with investigation findings
2. Present to operator: Show the proposed comment for confirmation
3. Add comment: Call jira_add_comment(issue_key, body) after confirmation

Example prompts:

• "Add a comment to NET-1234: 'Root cause identified - BGP hold timer expired due to high CPU'"
• "Update OPS-567 with: 'Fan replacement scheduled for maintenance window'"

Confluence Write Workflows (User Story 4)

Creating Pages

Create new Confluence pages with human-in-the-loop confirmation.

Step-by-step workflow:

1. Verify target space: Call confluence_get_space(space_key) to confirm space exists
2. Present to operator: Display proposed page title, space, and body content
3. Create page: After confirmation, call confluence_create_page(space_key, title, body)
4. Confirm creation: Return the created page URL to the operator

Example prompts:

• "Create a Confluence page in NET-DOCS titled 'Incident Postmortem: Router-01 Outage'"
• "Create a new runbook page for BGP troubleshooting in the network documentation space"

IMPORTANT: Always present the full page content to the operator and wait for explicit confirmation before calling confluence_create_page.

Updating Pages

Update existing page content (creates a new version).

Step-by-step workflow:

1. Get current page: Call confluence_get_page(page_id) to retrieve current content and version (read-before-write)
2. Present changes: Show operator the proposed changes (diff view if possible)
3. Update page: After confirmation, call confluence_update_page(page_id, title, body, version) - note: version must be incremented
4. Confirm update: Verify the new version was created

Example prompts:

• "Update the 'Router Maintenance Procedures' page with the new reboot procedure"
• "Add a section to the BGP runbook about peer flapping troubleshooting"

Adding Page Comments

Add comments to Confluence pages.

Step-by-step workflow:

1. Draft comment: Prepare comment text
2. Present to operator: Show the proposed comment for confirmation
3. Add comment: Call confluence_add_comment(page_id, body) after confirmation

Example prompts:

• "Add a comment to the postmortem page noting the action items are complete"

Deleting Pages

Delete Confluence pages (requires explicit confirmation).

Step-by-step workflow:

1. Get page details: Call confluence_get_page(page_id) to show what will be deleted
2. Explicit warning: Warn operator this is a destructive action
3. Delete page: Only after explicit confirmation, call confluence_delete_page(page_id)

IMPORTANT: Page deletion is destructive. Require explicit confirmation including the page title.

Cross-Product and Bulk Workflows (User Story 5)

Linking Jira Issues

Create relationships between issues.

Step-by-step workflow:

1. Get link types: Call jira_get_link_types to see available link types (Blocks, Relates, Clones, etc.)
2. Present to operator: Show the proposed link between issues
3. Create link: Call jira_link_issues(inward_issue, outward_issue, link_type) after confirmation
4. View links: Call jira_get_issue_links(issue_key) to verify the link was created

Example prompts:

• "Link NET-1234 as blocking OPS-567"
• "Show me all issues linked to NET-1234"
• "What link types are available in Jira?"

Bulk Issue Creation

Create multiple issues in a single batch operation.

Step-by-step workflow:

1. Prepare issue list: Collect all issue details (project, type, summary, description, priority)
2. Present full batch: Show operator ALL proposed issues for review
3. Create batch: After confirmation of the entire batch, call jira_batch_create_issues(issues)
4. Report results: Return all created issue keys

Example prompts:

• "Create 5 Jira issues in project NET for these network upgrade tasks"
• "Bulk create issues for each router in the maintenance list"

IMPORTANT: The entire batch must be confirmed before any issues are created.

Incident-to-Postmortem Workflow

Cross-product workflow: Jira incident with linked Confluence postmortem.

Step-by-step workflow:

1. Create Jira incident: Following the issue creation workflow
2. Create Confluence postmortem: Create postmortem page in appropriate space
3. Link documents: Add the Confluence page URL to the Jira issue via remote link or description update
4. Cross-reference: Add the Jira issue key to the Confluence page

Example prompts:

• "Create an incident in Jira for the router-01 outage and a postmortem page in Confluence"
• "Link Confluence postmortem page 123456 to Jira issue NET-1234"

GAIT Audit Logging

All Atlassian interactions must be logged to the GAIT audit trail via gait_mcp tools at skill invocation level.

Logging Pattern

For each operation, log:

1. Tool name: Which Atlassian MCP tool was called
2. Parameters: Key parameters (issue key, project, search query)
3. Result summary: Success/failure, created item keys, record counts

Example Log Entries

[GAIT] jira_search - JQL: "project = NET AND status != Done" - Found 23 issues
[GAIT] jira_get_issue - Issue: NET-1234 - Retrieved successfully
[GAIT] jira_create_issue - Project: NET, Type: Bug, Summary: "Router-01 BGP down" - Created: NET-1235
[GAIT] jira_transition_issue - Issue: NET-1234 - Transitioned: Open -> In Progress
[GAIT] confluence_create_page - Space: NET-DOCS, Title: "Postmortem: Router Outage" - Created: page_id=456789

Session Requirements

• Start every Atlassian session with gait_branch to create an audit branch
• Record every meaningful interaction with gait_record_turn
• End sessions with gait_log to display the full audit trail

Error Handling

Common Errors

| Error | Cause | Resolution | |-------|-------|------------| | Connection failure | Atlassian instance unreachable | Check JIRA_URL/CONFLUENCE_URL and network connectivity | | Authentication failure | Invalid token | Regenerate API token or PAT | | Not found | Issue/page/project doesn't exist | Verify the key/ID is correct | | Permission denied | Token lacks permissions | Check user's project/space permissions | | Rate limited | Too many requests | Wait and retry with backoff | | Invalid transition | Transition not available | Check jira_get_transitions for valid options | | Validation error | Missing required fields | Check required fields with jira_get_fields |

Graceful Degradation

• If only Jira is configured: Confluence tools are unavailable
• If only Confluence is configured: Jira tools are unavailable
• Always check tool availability before attempting operations

Safety Checklist

Before any write operation:

• ] Read current state first (Constitution II: Read-Before-Write)
• ] Present changes to operator for confirmation (Constitution XIV: Human-in-the-Loop)
• ] Log the operation to GAIT audit trail (Constitution III: GAIT Logging)
• ] Verify the operation succeeded before reporting completion

Example Workflows

Incident Response Workflow

1. Search for similar past incidents: jira_search with relevant JQL
2. Create new incident issue: jira_create_issue (with confirmation)
3. Search for relevant runbook: confluence_search
4. Transition to In Progress: jira_transition_issue (with confirmation)
5. Add investigation updates: jira_add_comment (with confirmation)
6. Resolve and document: Update issue and create postmortem page

Knowledge Base Maintenance

1. Search for outdated pages: confluence_search with date filters
2. Review page content: confluence_get_page
3. Update with current information: confluence_update_page (with confirmation)
4. Create related Jira task if major update needed: jira_create_issue