菜单
Use when the user wants to create a Linear ticket, issue, user story, or task - interviews the user adaptively, formats as a user story with test plan, and creates it via Linear MCP after confirmation
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/dkuehlwein/nova --skill create-ticket命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use when the user is done working on a Linear ticket and wants to wrap up - creates PR, runs code review, merges, updates the ticket status, and cleans up the worktree
Use when the user wants to start working on a Linear ticket - fetches ticket details, creates a git worktree with a properly named branch, and launches a subagent in the worktree with full ticket context and the right workflow skill
正在显示 SKILL.md
基于 SOC 职业分类
| name | create-ticket |
| description | Use when the user wants to create a Linear ticket, issue, user story, or task - interviews the user adaptively, formats as a user story with test plan, and creates it via Linear MCP after confirmation |
Interview the user to understand their request, then create a well-structured Linear ticket as a user story with acceptance criteria and a test plan. Tickets follow TDD: they're only Done when all tests pass.
Key principle: A ticket documents the what (the problem) and how to verify (the test plan), never the how to fix. Root cause analysis, code investigation, and solution design belong to the implementation phase when someone picks up the ticket. Code changes between ticket creation and implementation, so any fix proposed at creation time risks being stale or misleading. Do NOT explore the codebase, search for root causes, or propose solutions during ticket creation.
Start by asking what the user wants to build or fix. Use AskUserQuestion to gather information adaptively:
Always ask first:
Then adapt based on complexity:
For simple/clear requests, propose sensible defaults and move to confirmation.
For complex/ambiguous requests, follow up with:
Guidelines:
Before drafting, assess whether the request is a single ticket or needs to be split. A good ticket should be completable in one focused effort.
Signs a ticket is too big:
If it's too big, propose splitting it. Present the breakdown to the user as a set of smaller tickets, each with its own user story. Ask which ones to create. Each sub-ticket should be independently valuable and testable - not just arbitrary slices.
If it's borderline, ask the user: "This could be one ticket or split into N. Preference?"
Format the ticket as a user story with this structure:
## User Story
As a [role], I want [goal], so that [benefit].
## Context
[Brief problem description or background - what's happening now and why it needs to change]
## Acceptance Criteria
- [ ] [Specific, testable criterion 1]
- [ ] [Specific, testable criterion 2]
- [ ] [...]
## Test Plan
### Unit Tests
- [ ] [Test case 1: what to test and expected outcome]
- [ ] [Test case 2: ...]
### Integration Tests (if applicable)
- [ ] [Test case: ...]
### Manual Verification (if applicable)
- [ ] [Step to verify: ...]
## Definition of Done
- [ ] All acceptance criteria met
- [ ] All test cases implemented and passing
- [ ] Code reviewed
Writing guidelines:
Present the full draft to the user in a formatted message. Ask for approval or changes.
On approval, create the issue using the Linear MCP tools:
On changes, revise and re-present.
When creating multiple tickets (from a split), create them all after approval. If there's a natural ordering, set up blocking relationships between them.
After creation, show the user the issue identifier(s) (e.g., NOV-123) so they can find them in Linear.
| Label | Use When |
|---|---|
| Bug | Something is broken or behaving incorrectly |
| Feature | Net-new functionality |
| Improvement | Enhancing existing functionality |
| Agent | Related to the AI agent system (backend/agent/) |
| Frontend | Related to the Next.js frontend |
| Tools | Related to MCP tools or LangChain tools |