Jira and Confluence via MCP: Queries and Safe Updates

Correct JQL/CQL, previewed Jira and Confluence writes, evidence-based sprint moves, and MCP auth fixes for Atlassian.

by AgentVoltv1.0.0
Free to sign up · every skill included with AgentVolt Pro

Before & after

Same prompt, same agent — the only difference is the skill.

Prompt“Pull all open tickets with the 'blocked' label.”
Beforegeneric AI agent
I can help you query your Jira instance for those tickets. To do this properly, I'd need to connect to your Jira workspace and run a JQL query. Do you want to filter by project as well, or just pull all blocked tickets across the whole workspace? Are you interested in any other filters like priority or assignee?
Afterwith Jira and Confluence via MCP: Queries and Safe Updates
JIRA QUERY RESULTS

QUERY: label = Blocked AND status != Done
RESULT COUNT: 27 open tickets

TICKETS BY PROJECT
Backend Services: 12 tickets
Frontend Web: 9 tickets
Mobile App: 4 tickets
Infrastructure: 2 tickets

TOP 3 BLOCKING ISSUES
1. AUTH-487 – "Implement SAML SSO"
   Blocked by: Platform decision on IdP vendor
   Priority: High
   Assignee: Jane Smith
   Days Blocked: 14

2. API-234 – "Rate limiting for v2 endpoints"
   Blocked by: AUTH-487 (dependency)
   Priority: Medium
   Assignee: Unassigned
   Days Blocked: 8

3. WEB-156 – "Dashboard performance optimization"
   Blocked by: API-234 (needs stable rate-limit API)
   Priority: Medium
   Assignee: John Lee
   Days Blocked: 5

BLOCKED DURATION DISTRIBUTION
Blocked 1-5 days: 8 tickets (new, resolvable)
Blocked 6-14 days: 12 tickets (investigation phase)
Blocked 15+ days: 7 tickets (stalled, needs escalation)

ACTION ITEMS
Immediate: Escalate 7 long-stalled tickets to team leads
This week: Schedule sync on AUTH-487 IdP decision
Next week: Unblock API-234 once AUTH-487 settles

SAVED FILTER: Open Blocked Tickets by Duration (reuse anytime)
JQL: label = Blocked AND status != Done ORDER BY created DESC

About this skill

Your Jira query misses half the tickets. A create call fails with "field is required." A Confluence edit wipes out every macro on the page. The MCP server returns 401 and nobody knows why. This skill makes Claude handle Jira and Confluence the way an experienced Atlassian power user does. It writes queries that return what you meant, looks up field IDs and transitions before writing anything, and shows you every bulk change before it happens.

What it does

  • Writes JQL that avoids the common silent misses: != dropping empty fields, hard-coded status names across workflows, and display-name user matching
  • Looks up required fields, allowed values, and transition IDs before creating or moving issues, so writes succeed on the first try
  • Estimates sprint headroom from recent velocity, because Jira stores no capacity field, and proposes which issues fit
  • Edits Confluence pages with version handling and storage-format care, so macros, mentions, and Jira embeds survive
  • Diagnoses MCP and API failures by status code (401 vs 403 vs 404 vs 429) and deployment type (Cloud vs Data Center)
  • Lists bulk changes as before-and-after rows and waits for your OK before writing

What you get

The exact query with a one-line reason for each non-obvious choice, the results, and a change plan you can confirm. After a write, you get a per-issue report of what changed (key, field, old value, new value) and anything that was skipped.

Best for

Engineers, tech leads, and scrum masters who run Jira and Confluence from Claude or Claude Code through an Atlassian MCP server and want correct queries and no surprise writes.

SKILL.md preview

SKILL.md
---
name: jira-confluence-mcp-queries-safe-updates

… (sign up to view the full skill)
Sign up to view, copy, and install the full skill

More development skills

View all Development skills →