jira-skill MCP — ユースケース, インストール & ライブデモ

なぜ使うのか

主な機能

Works with Jira Server, DC, and Cloud
No Docker — pure shell + Python
Full issue lifecycle: create, transition, comment, link, worklog, attach
JQL search helpers
Sprint + board management
Interactive jira-setup.py credential configurator
jira-syntax helper for wiki markup

ライブデモ

実際の動作

jira-skill.replay ▶ 準備完了

0/0

インストール

クライアントを選択

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "jira-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/netresearch/jira-skill",
        "~/.claude/skills/jira-skill"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config を開く。保存後、アプリを再起動。

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "jira-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/netresearch/jira-skill",
        "~/.claude/skills/jira-skill"
      ],
      "_inferred": true
    }
  }
}

Cursor は Claude Desktop と同じ mcpServers スキーマを使用。プロジェクト設定はグローバルより優先。

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "jira-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/netresearch/jira-skill",
        "~/.claude/skills/jira-skill"
      ],
      "_inferred": true
    }
  }
}

Cline サイドバーの MCP Servers アイコンをクリックし、"Edit Configuration" を選択。

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "jira-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/netresearch/jira-skill",
        "~/.claude/skills/jira-skill"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop と同じ形式。Windsurf を再起動して反映。

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "jira-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/netresearch/jira-skill",
        "~/.claude/skills/jira-skill"
      ]
    }
  ]
}

Continue はマップではなくサーバーオブジェクトの配列を使用。

~/.config/zed/settings.json

{
  "context_servers": {
    "jira-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/netresearch/jira-skill",
          "~/.claude/skills/jira-skill"
        ]
      }
    }
  }
}

context_servers に追加。保存時に Zed がホットリロード。

claude mcp add jira-skill -- git clone https://github.com/netresearch/jira-skill ~/.claude/skills/jira-skill

ワンライナー。claude mcp list で確認、claude mcp remove で削除。

ユースケース

実用的な使い方： jira-skill

Triage a sprint backlog and unstick blocked issues

👤 Scrum masters, tech leads ⏱ ~20 min beginner

使うタイミング： Mid-sprint when velocity is dropping and you suspect blockers.

前提条件

Jira credentials configured — Run jira-setup.py and provide URL + email + API token (Cloud) or PAT (Server/DC)
Skill installed — git clone https://github.com/netresearch/jira-skill ~/.claude/skills/jira-skill

フロー

List sprint issues by status

Use jira-skill. List issues in sprint 'Sprint 42' grouped by status. Flag anything not updated in 3 days.✓ コピーしました

→ Grouped table with staleness markers
Drill into blockers

For the blocked issues, pull the last comment and linked issues — who's waiting on what?✓ コピーしました

→ Dependency summary
Nudge owners

Draft a comment on each blocked issue pinging the responsible party.✓ コピーしました

→ Per-issue drafts

結果： A clear list of unblock actions to drive in standup.

注意点

Over-pinging — comments cost attention — Let Claude draft; post manually only the ones that matter

組み合わせ： github

Generate release notes from closed issues

👤 Release managers ⏱ ~15 min beginner

使うタイミング： Cutting a release and need user-facing notes from the issue list.

フロー

JQL for the release

JQL: fixVersion = 1.42 AND status = Done. Pull summaries and types.✓ コピーしました

→ Ordered issue list
Draft notes

Group by type (Feature / Bug / Improvement) and write user-facing release notes — no internal jargon.✓ コピーしました

→ Markdown notes ready to publish

結果： Release notes in minutes instead of hours.

組み合わせ： github

Bulk-transition issues through a workflow

👤 Project admins ⏱ ~10 min intermediate

使うタイミング： Moving a batch of issues between statuses (e.g. deploy-complete).

フロー

Dry-run first

Find all 'In Staging' issues with fixVersion = 1.42. Don't transition yet — just list.✓ コピーしました

→ List to confirm
Transition

Transition those to 'Done' with a comment 'Deployed to production in 1.42'.✓ コピーしました

→ Per-issue transition result

結果： Clean workflow state, nothing left in 'In Staging'.

注意点

Missing required fields on transition — Skill surfaces the required field; supply it in the prompt

Summarize a team's worklog for the month

👤 Engineering managers, PMs tracking capacity ⏱ ~20 min intermediate

使うタイミング： Month-end capacity review.

フロー

Query worklogs

Pull all worklogs for team members in project ACME for last month.✓ コピーしました

→ Per-person, per-issue hour totals
Summarize

Roll up to epic and share of total.✓ コピーしました

→ Top-level breakdown

結果： Capacity picture grounded in actual logged work.

注意点

Worklogs aren't filled in — bad data — Flag people with 0 logs; don't draw conclusions from missing data

組み合わせ

他のMCPと組み合わせて10倍の力を

jira-skill + github

Link PRs to Jira issues or create Jira issues from GitHub bugs

For each GitHub issue labeled 'bug' opened this week, create a Jira ticket in project ACME.✓ コピーしました

ツール

このMCPが提供する機能

ツール	入力	呼び出すタイミング	コスト
get_issue	issue key	Anytime	1 API call
search_jql	JQL string	Discovery	1 API call
create_issue	project, type, summary, description	Filing new issue	1 API call (write)
transition_issue	key, transition id/name	Workflow moves	1 API call (write)
add_comment	key, body (jira-syntax ok)	Nudges + notes	1 API call (write)
link_issues	key, inward, link type	Dependencies	1 API call
add_worklog	key, time, comment	Time tracking	1 API call
sprint_manage	board, sprint, action	Scrum ops	1 API call

コストと制限

運用コスト

APIクォータ: Jira Cloud: 10 req/sec per user typical; Server/DC is instance-configured
呼び出しあたりのトークン: Low — mostly API responses
金額: Free skill; Jira license separate
ヒント: Use JQL server-side filters rather than pulling all and filtering client-side

セキュリティ

権限、シークレット、影響範囲

最小スコープ： Read project; write only where needed (transition/comment)

認証情報の保管： JIRA_URL + JIRA_USER + JIRA_TOKEN in env; don't check into git

データ送信先： Only to your Jira host

絶対に付与しない： Admin on the instance

API tokens for Cloud should be user-scoped; for Server/DC use personal access tokens, not passwords.

トラブルシューティング

よくあるエラーと対処法

401 Unauthorized

Token missing or revoked; re-run jira-setup.py

確認： curl -u $JIRA_USER:$JIRA_TOKEN $JIRA_URL/rest/api/2/myself

400 on transition

Transition requires a field you didn't provide — inspect transitions with get_transitions

Search returns 0 but issues exist

Your JQL or project permissions; verify the user can see the project

代替案

jira-skill 他との比較

代替案	代わりに使う場面	トレードオフ
Atlassian's official MCP	You prefer first-party integration	Less control over shell access

その他

リソース

📖 GitHub の公式 README を読む

🐙 オープンな issue を見る

🔍 400以上のMCPサーバーとSkillsを見る