mk:jira — Routing Skill

What This Skill Does

mk:jira is a pure routing layer as of v2.8.3. Its sole purpose is to identify the right mk:jira-* leaf for the user's Jira task and point at it. No execution.

The Atlassian-MCP-coupled implementation has been removed. Jira execution now lives in 16 thin leaf skills (each forks a dedicated agent in .claude/agents/) backed by the jira-as CLI installed via npx mewkit setup.

When to Use

• Triggers: "jira", "jira ticket", ambiguous Jira intent
• NOT for: Direct execution — forward to a specific mk:jira-{leaf}. If you already know the leaf, invoke it directly.

Routing Table

User intent	Leaf skill
Get/create/update/delete a single issue	mk:jira-issue
Find issues by criteria; export; filters	mk:jira-search
Transition through workflow; assign; resolve	mk:jira-lifecycle
Comments, attachments, watchers, notify	mk:jira-collaborate
Issue links, blockers, dependencies	mk:jira-relationships
Worklogs, time tracking, reports	mk:jira-time
Epics, sprints, backlog, story points	mk:jira-agile
Custom field discovery + configuration	mk:jira-fields
Bulk ops (10+ issues, dry-run mandatory)	mk:jira-bulk
Service Management (queues, SLAs, customers)	mk:jira-jsm
Project/user/group/scheme administration	mk:jira-admin
Git/PR/branch-name integration	mk:jira-dev
Cache + project-context discovery	mk:jira-ops
Evaluate ticket complexity + inconsistencies	mk:jira-evaluator
Estimate story points (heuristic)	mk:jira-estimator
Analyze ticket context + media; RCA	mk:jira-analyst

Setup

One-time:

npx mewkit setup           # auto-installs jira-as into .claude/skills/.venv
cp .claude/.env.example .claude/.env
# edit MEOW_JIRA_API_TOKEN, MEOW_JIRA_EMAIL, MEOW_JIRA_SITE_URL

The SessionStart hook jira-env-loader.sh validates .claude/.env and emits [mk:jira] env OK / [mk:jira] <KEY> missing. The wrapper scripts/jira-as.sh translates MEOW_JIRA_* → native JIRA_* env per call.

See Environment Variables — Jira for the full list.

Architecture

mk:jira (router, ≤80 lines)              ← recommends a leaf, never executes
├── shared resources (used by all 16 leaves):
│   ├── scripts/{jira-as.sh, jql-sanitize.sh, fetch-workflow.sh}
│   ├── scripts/requirements.txt          ← pip deps installed by mewkit setup
│   └── references/{install-and-auth, cli-idioms, safety-framework}.md
└── 16 thin leaf skills (each: context: fork → matching agent in .claude/agents/)
    ├── 13 domain leaves (issue, search, lifecycle, collaborate, relationships, time,
    │                     agile, fields, bulk, jsm, admin, dev, ops)
    └── 3 intelligence leaves (evaluator, estimator, analyst)

tasks/jira-workflows/              ← discovered Jira workflow cache (NEW in 2.8.3)
├── README.md
├── _schemes/<PROJECT_KEY>.md      ← project → workflow-name mapping
└── <workflow-slug>.md             ← full statuses + transitions per workflow

Handoff

• mk:intake → recommends the right mk:jira-* leaf based on triage
• mk:planning-engine → consumes mk:jira-evaluator + mk:jira-estimator reports for capacity analysis
• mk:cook Phase 3 → may invoke mk:jira-issue to create implementation tickets
• mk:ship Phase 5 → may invoke mk:jira-dev for branch-name + PR description

Gotchas

• mk:jira itself never executes; if you find yourself thinking about jira-as commands inside this skill, you're in the wrong place — forward to a leaf.
• Escape hatch: if jira-as is unusable (mTLS, multi-profile), invoke Atlassian MCP directly per references/install-and-auth.md. The router does NOT auto-fallback.