本文へ移動
cccskills
無料GitHub で公開

jira-workflow

Orchestrate Jira workflows end-to-end. Use when building stories with approvals, transitioning items through lifecycle states, or syncing task completion with Jira.

インストール方法を見る

含まれるファイル(2)

  • SKILL.md18.8 KB
  • skill-report.json109.1 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Jira Workflow Orchestration Skill

Complete workflow management for Jira: building stories (SAFe), getting approvals, and transitioning items through the development lifecycle (To Do → Progressing → Done).

IMPORTANT: This project uses Next-Gen (Team-managed) Jira with custom workflow states. The actual states are:

  • To Do (backlog)
  • In Review
  • Progressing (active work)
  • Out Review
  • Done

Always query available transitions first: GET /rest/api/3/issue/{key}/transitions

When to Use

  • Creating new user stories, epics, or tasks for the project
  • Getting user approval before creating Jira items
  • Moving stories through workflow states as work progresses
  • Syncing Claude Code task completion with Jira status
  • Managing sprint planning and backlog refinement
  • Tracking development progress in real-time

Prerequisites

Environment Variables:

JIRA_EMAIL=your.email@domain.com
JIRA_API_TOKEN=your_api_token
JIRA_BASE_URL=https://your-org.atlassian.net
JIRA_PROJECT_KEY=SCRUM
JIRA_BOARD_ID=1

Project Configuration:

  • Must know if project is Next-Gen (Team-managed) or Classic (Company-managed)
  • Next-Gen: Use parent field for Epic links
  • Classic: Use customfield_10014 for Epic links

Core Workflow Pattern

The Approval-Create-Track Loop

1. PLAN: Analyze task requirements
   ↓
2. PROPOSE: Present story to user for approval
   ↓
3. APPROVE: User confirms or modifies
   ↓
4. CREATE: Issue created in Jira backlog
   ↓
5. START: Transition to "Progressing" when work begins
   ↓
6. COMPLETE: Transition to "Done" when work verified
   ↓
7. SYNC: Update Jira with implementation details

Phase 1: Story Building (SAFe Format)

Building a Story Proposal

When user requests work, build a SAFe-compliant story proposal:

function buildStoryProposal(task) {
  return {
    summary: `As a ${task.persona}, I want ${task.goal}, so that ${task.benefit}`,
    description: {
      userStory: `As a **${task.persona}**, I want **${task.goal}**, so that **${task.benefit}**.`,
      acceptanceCriteria: task.scenarios.map(s => ({
        name: s.name,
        given: s.given,
        when: s.when,
        then: s.then
      })),
      definitionOfDone: [
        'Code reviewed and approved',
        'Unit tests written and passing',
        'Integration tests passing',
        'Documentation updated',
        'Deployed to staging',
        'Validated in production'
      ],
      technicalNotes: task.technicalNotes || []
    },
    category: task.category, // authentication, ui, api, database, etc.
    estimatedComplexity: task.complexity || 'medium', // small, medium, large
    subtasks: task.subtasks || []
  };
}

Presenting for Approval

CRITICAL: Always get user approval before creating Jira items.

Use this prompt pattern:

## Proposed Jira Story

**Summary:** As a [persona], I want [goal], so that [benefit]

**Category:** [category]
**Complexity:** [small/medium/large]

### Acceptance Criteria

**Scenario 1: [Name]**
- **GIVEN** [precondition]
- **WHEN** [action]
- **THEN** [expected result]

### Subtasks (if any)
1. [Subtask 1]
2. [Subtask 2]
3. [Subtask 3]

---

**Do you want me to create this in Jira?**

Options:
1. **Yes, create as-is** - I'll create the story now
2. **Modify** - Tell me what to change
3. **Skip** - Don't create in Jira, just do the work

Phase 2: Issue Creation

Create Story in Jira

const JIRA_EMAIL = process.env.JIRA_EMAIL;
const JIRA_API_TOKEN = process.env.JIRA_API_TOKEN;
const JIRA_BASE_URL = process.env.JIRA_BASE_URL;
const PROJECT_KEY = process.env.JIRA_PROJECT_KEY;

const auth = Buffer.from(`${JIRA_EMAIL}:${JIRA_API_TOKEN}`).toString('base64');
const headers = {
  'Authorization': `Basic ${auth}`,
  'Content-Type': 'application/json',
  'Accept': 'application/json'
};

async function createStory(proposal, epicKey = null) {
  const body = {
    fields: {
      project: { key: PROJECT_KEY },
      issuetype: { name: 'Story' },
      summary: proposal.summary,
      description: buildADF(proposal.description),
      labels: [proposal.category.toLowerCase().replace(/\s+/g, '-')]
    }
  };

  // Link to Epic (Next-Gen project)
  if (epicKey) {
    body.fields.parent = { key: epicKey };
  }

  const response = await fetch(`${JIRA_BASE_URL}/rest/api/3/issue`, {
    method: 'POST',
    headers,
    body: JSON.stringify(body)
  });

  if (!response.ok) {
    const error = await response.text();
    throw new Error(`Failed to create story: ${error}`);
  }

  const issue = await response.json();
  console.log(`Created: ${issue.key} - ${proposal.summary}`);

  // Create subtasks if any
  if (proposal.subtasks?.length > 0) {
    for (const subtask of proposal.subtasks) {
      await createSubtask(issue.key, subtask);
      await delay(100); // Rate limiting
    }
  }

  return issue;
}

async function createSubtask(parentKey, summary) {
  const body = {
    fields: {
      project: { key: PROJECT_KEY },
      issuetype: { name: 'Subtask' }, // Note: 'Subtask' for Next-Gen
      parent: { key: parentKey },
      summary: summary
    }
  };

  const response = await fetch(`${JIRA_BASE_URL}/rest/api/3/issue`, {
    method: 'POST',
    headers,
    body: JSON.stringify(body)
  });

  if (!response.ok) {
    const error = await response.text();
    throw new Error(`Failed to create subtask: ${error}`);
  }

  return response.json();
}

function delay(ms) {
  return new Promise(resolve => setTimeout(resolve, ms));
}

Build Atlassian Document Format (ADF)

function buildADF(content) {
  const sections = [];

  // User Story Section
  sections.push({
    type: 'heading',
    attrs: { level: 2 },
    content: [{ type: 'text', text: 'User Story' }]
  });
  sections.push({
    type: 'paragraph',
    content: [{ type: 'text', text: content.userStory }]
  });

  // Acceptance Criteria Section
  sections.push({
    type: 'heading',
    attrs: { level: 2 },
    content: [{ type: 'text', text: 'Acceptance Criteria' }]
  });

  for (const scenario of content.acceptanceCriteria) {
    sections.push({
      type: 'heading',
      attrs: { level: 3 },
      content: [{ type: 'text', text: `Scenario: ${scenario.name}` }]
    });
    sections.push({
      type: 'bulletList',
      content: [
        { type: 'listItem', content: [{ type: 'paragraph', content: [{ type: 'text', text: `GIVEN ${scenario.given}`, marks: [{ type: 'strong' }] }] }] },
        { type: 'listItem', content: [{ type: 'paragraph', content: [{ type: 'text', text: `WHEN ${scenario.when}`, marks: [{ type: 'strong' }] }] }] },
        { type: 'listItem', content: [{ type: 'paragraph', content: [{ type: 'text', text: `THEN ${scenario.then}`, marks: [{ type: 'strong' }] }] }] }
      ]
    });
  }

  // Definition of Done Section
  sections.push({
    type: 'heading',
    attrs: { level: 2 },
    content: [{ type: 'text', text: 'Definition of Done' }]
  });
  sections.push({
    type: 'bulletList',
    content: content.definitionOfDone.map(item => ({
      type: 'listItem',
      content: [{ type: 'paragraph', content: [{ type: 'text', text: `[ ] ${item}` }] }]
    }))
  });

  // Technical Notes (if any)
  if (content.technicalNotes?.length > 0) {
    sections.push({
      type: 'heading',
      attrs: { level: 2 },
      content: [{ type: 'text', text: 'Technical Notes' }]
    });
    sections.push({
      type: 'bulletList',
      content: content.technicalNotes.map(note => ({
        type: 'listItem',
        content: [{ type: 'paragraph', content: [{ type: 'text', text: note }] }]
      }))
    });
  }

  return { type: 'doc', version: 1, content: sections };
}

Phase 3: Workflow Transitions

Get Available Transitions

async function getTransitions(issueKey) {
  const response = await fetch(
    `${JIRA_BASE_URL}/rest/api/3/issue/${issueKey}/transitions`,
    { headers }
  );

  if (!response.ok) {
    throw new Error(`Failed to get transitions: ${response.status}`);
  }

  const data = await response.json();
  return data.transitions;
}

Transition Issue to State

async function transitionTo(issueKey, targetState) {
  // Get available transitions
  const transitions = await getTransitions(issueKey);

  // Find the transition to target state
  const transition = transitions.find(t =>
    t.to.name.toLowerCase() === targetState.toLowerCase() ||
    t.name.toLowerCase() === targetState.toLowerCase()
  );

  if (!transition) {
    console.log(`Available transitions for ${issueKey}:`);
    transitions.forEach(t => console.log(`  - ${t.name} → ${t.to.name}`));
    throw new Error(`No transition to "${targetState}" found`);
  }

  // Execute the transition
  const response = await fetch(
    `${JIRA_BASE_URL}/rest/api/3/issue/${issueKey}/transitions`,
    {
      method: 'POST',
      headers,
      body: JSON.stringify({ transition: { id: transition.id } })
    }
  );

  if (!response.ok) {
    const error = await response.text();
    throw new Error(`Failed to transition: ${error}`);
  }

  console.log(`${issueKey} transitioned to ${targetState}`);
  return true;
}

Common Workflow Operations

// Start work on a story (To Do → Progressing)
async function startWork(issueKey) {
  await transitionTo(issueKey, 'Progressing');
  console.log(`Started: ${issueKey}`);
}

// Complete a story (Progressing → Done)
async function completeWork(issueKey) {
  await transitionTo(issueKey, 'Done');
  console.log(`Completed: ${issueKey}`);
}

// Move back to backlog (any state → To Do)
async function moveToBacklog(issueKey) {
  await transitionTo(issueKey, 'To Do');
  console.log(`Moved to backlog: ${issueKey}`);
}

// Reopen a completed issue (Done → To Do)
async function reopenWork(issueKey) {
  await transitionTo(issueKey, 'To Do');
  console.log(`Reopened: ${issueKey}`);
}

Phase 4: Add Comments and Updates

Add Work Log Comment

async function addComment(issueKey, comment) {
  const body = {
    body: {
      type: 'doc',
      version: 1,
      content: [
        {
          type: 'paragraph',
          content: [{ type: 'text', text: comment }]
        }
      ]
    }
  };

  const response = await fetch(
    `${JIRA_BASE_URL}/rest/api/3/issue/${issueKey}/comment`,
    {
      method: 'POST',
      headers,
      body: JSON.stringify(body)
    }
  );

  if (!response.ok) {
    throw new Error(`Failed to add comment: ${response.status}`);
  }

  console.log(`Comment added to ${issueKey}`);
  return response.json();
}

Add Implementation Details Comment

async function addImplementationDetails(issueKey, details) {
  const content = [
    { type: 'heading', attrs: { level: 3 }, content: [{ type: 'text', text: 'Implementation Details' }] },
    { type: 'paragraph', content: [{ type: 'text', text: `Completed: ${new Date().toISOString()}` }] }
  ];

  if (details.files?.length > 0) {
    content.push(
      { type: 'heading', attrs: { level: 4 }, content: [{ type: 'text', text: 'Files Modified' }] },
      {
        type: 'bulletList',
        content: details.files.map(f => ({
          type: 'listItem',
          content: [{ type: 'paragraph', content: [{ type: 'text', text: f }] }]
        }))
      }
    );
  }

  if (details.commits?.length > 0) {
    content.push(
      { type: 'heading', attrs: { level: 4 }, content: [{ type: 'text', text: 'Commits' }] },
      {
        type: 'bulletList',
        content: details.commits.map(c => ({
          type: 'listItem',
          content: [{ type: 'paragraph', content: [{ type: 'text', text: c }] }]
        }))
      }
    );
  }

  if (details.notes) {
    content.push(
      { type: 'heading', attrs: { level: 4 }, content: [{ type: 'text', text: 'Notes' }] },
      { type: 'paragraph', content: [{ type: 'text', text: details.notes }] }
    );
  }

  const body = { body: { type: 'doc', version: 1, content } };

  const response = await fetch(
    `${JIRA_BASE_URL}/rest/api/3/issue/${issueKey}/comment`,
    {
      method: 'POST',
      headers,
      body: JSON.stringify(body)
    }
  );

  return response.json();
}

Complete Workflow Example

Full Cycle: Propose → Approve → Create → Work → Complete

async function fullWorkflowCycle(task) {
  // 1. Build proposal
  const proposal = buildStoryProposal(task);

  // 2. Present for approval (use AskUserQuestion tool)
  const approved = await presentForApproval(proposal);

  if (!approved) {
    console.log('Story creation skipped by user');
    return null;
  }

  // 3. Create in Jira
  const issue = await createStory(proposal, task.epicKey);
  console.log(`Created: ${issue.key}`);

  // 4. Start work (transition to In Progress)
  await startWork(issue.key);

  // 5. Do the actual work (your implementation here)
  const result = await doTheWork(task);

  // 6. Add implementation details
  await addImplementationDetails(issue.key, {
    files: result.modifiedFiles,
    commits: result.commits,
    notes: result.notes
  });

  // 7. Complete the work
  await completeWork(issue.key);

  return issue;
}

Integration with Claude Code Orchestration

Sync with TodoWrite

When working on Jira stories, sync with TodoWrite:

TodoWrite todos:
[
  { "content": "SCRUM-55: Create signup API", "status": "in_progress", "activeForm": "Working on SCRUM-55" },
  { "content": "SCRUM-56: Create login API", "status": "pending", "activeForm": "Waiting for SCRUM-55" },
  { "content": "SCRUM-57: Create logout API", "status": "pending", "activeForm": "Waiting for SCRUM-56" }
]

As each task completes:
1. Mark TodoWrite item as completed
2. Transition Jira issue to Done
3. Add implementation comment to Jira
4. Move to next task

Auto-Transition Pattern

// When starting a task
async function startTask(issueKey) {
  // 1. Transition Jira to Progressing
  await startWork(issueKey);

  // 2. Update TodoWrite (in Claude Code)
  // TodoWrite: Mark as in_progress

  return issueKey;
}

// When completing a task
async function completeTask(issueKey, details) {
  // 1. Add implementation comment
  await addImplementationDetails(issueKey, details);

  // 2. Transition Jira to Done
  await completeWork(issueKey);

  // 3. Update TodoWrite (in Claude Code)
  // TodoWrite: Mark as completed

  return issueKey;
}

Quick Reference

Status Transitions (SCRUM Project - Next-Gen)

FromToTransition NameTypical Use
To DoProgressing"Progressing"Starting work
To DoIn Review"In Review"Needs review first
ProgressingDone"Done"Work complete
ProgressingTo Do"To Do"Blocked/deprioritized
DoneTo Do"To Do"Reopening

Available States: To Do, In Review, Progressing, Out Review, Done

Note: Always query transitions first - they vary by issue type and current state.

API Endpoints

ActionMethodEndpoint
Create IssuePOST/rest/api/3/issue
Get IssueGET/rest/api/3/issue/{key}
Update IssuePUT/rest/api/3/issue/{key}
Delete IssueDELETE/rest/api/3/issue/{key}
Get TransitionsGET/rest/api/3/issue/{key}/transitions
Do TransitionPOST/rest/api/3/issue/{key}/transitions
Add CommentPOST/rest/api/3/issue/{key}/comment
SearchGET/rest/api/3/search/jql?jql=...

Rate Limiting

  • Max 10 requests/second
  • Add 100ms delay between bulk operations
  • Batch operations where possible

Error Handling

async function safeJiraOperation(operation, issueKey) {
  try {
    return await operation();
  } catch (error) {
    console.error(`Jira operation failed for ${issueKey}: ${error.message}`);

    // Common error patterns
    if (error.message.includes('404')) {
      console.log('Issue not found - may have been deleted');
    }
    if (error.message.includes('401')) {
      console.log('Authentication failed - check API token');
    }
    if (error.message.includes('403')) {
      console.log('Permission denied - check project access');
    }
    if (error.message.includes('400')) {
      console.log('Bad request - check field names and values');
    }

    throw error;
  }
}

Executable Scripts

Ready-to-run scripts are available in both Node.js and Python:

Using the Cross-Platform Runner

# From the .claude/skills/jira directory
node scripts/run.js workflow demo SCRUM-100  # Demo full workflow
node scripts/run.js test                      # Test authentication

# Force specific runtime
node scripts/run.js --python workflow demo SCRUM-100
node scripts/run.js --node workflow demo SCRUM-100

Direct Script Execution

# Node.js
node scripts/jira-workflow-demo.mjs demo SCRUM-100
node scripts/jira-workflow-demo.mjs start SCRUM-100
node scripts/jira-workflow-demo.mjs complete SCRUM-100
node scripts/jira-workflow-demo.mjs reopen SCRUM-100
node scripts/jira-workflow-demo.mjs status SCRUM-100

# Python (recommended on Windows)
python scripts/jira-workflow-demo.py demo SCRUM-100
python scripts/jira-workflow-demo.py start SCRUM-100
python scripts/jira-workflow-demo.py complete SCRUM-100
python scripts/jira-workflow-demo.py reopen SCRUM-100
python scripts/jira-workflow-demo.py status SCRUM-100

Available Scripts

ScriptNode.jsPythonPurpose
Workflow Demojira-workflow-demo.mjsjira-workflow-demo.pyFull To Do → Progressing → Done demo
Add Subtasksjira-add-subtasks.mjsjira-add-subtasks.pyCreate subtasks under a story
Create Storyjira-create-one.mjsjira-create-one.pyCreate single story
Bulk Createjira-bulk-create.mjsjira-bulk-create.pyCreate from git commits

References

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Trigger for: explaining current FAR, DFARS, or agency-supplement text; determining documented acquisition-policy status for an agency and FAR part; comparing codified text, RFO model text, and agency deviations; tracing acquisition rulemaking; finding procurement comment periods; analyzing public comments; refreshing prior policy analysis; or producing an Acquisition Policy Impact Brief. Use published federal sources and supplied policy documents without making legal or procurement-specific applicability determinations. Do not use for clause selection, market research execution, opportunities, pricing, grants, or cooperative agreements.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Systematic framework for extracting actionable items from documents and organizing them into prioritized, trackable checklists. Use when converting reports, meeting notes, audits, or any document with embedded action items into structured TODO lists.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Browser automation for AI agents via inference.sh. Navigate web pages, interact with elements using @e refs, take screenshots, record video. Capabilities: web scraping, form filling, clicking, typing, drag-drop, file upload, JavaScript execution. Use for: web automation, data extraction, testing, agent browsing, research. Triggers: browser, web automation, scrape, navigate, click, fill form, screenshot, browse web, playwright, headless browser, web agent, surf internet, record video

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Run AI apps via inference.sh CLI - image generation, video creation, LLMs, search, 3D, Twitter automation. Models: FLUX, Veo, Gemini, Grok, Claude, Seedance, OmniHuman, Tavily, Exa, OpenRouter, and many more. Use when running AI apps, generating images/videos, calling LLMs, web search, or automating Twitter. Triggers: inference.sh, infsh, ai model, run ai, serverless ai, ai api, flux, veo, claude api, image generation, video generation, openrouter, tavily, exa search, twitter api, grok

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

Agile ceremonies and sprint management specialist. Use when running standups, retrospectives, sprint planning, or tracking sprint progress.

日本語の概要は準備中です。原文の説明を表示しています。

aiskillstore/marketplace4322026年10月13日 更新

aiskillstore のスキルをすべて見る

このスキルの問題を報告する