메뉴
Create a Linear issue with consistent structure, inheriting context from related issues when possible
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | create-linear-issue |
| description | Create a Linear issue with consistent structure, inheriting context from related issues when possible |
Create a Linear issue from the current conversation. Triggers on "create a Linear issue", "file a ticket", "open a Linear issue", "make a Linear issue for X".
The Linear MCP tools are deferred and must be loaded before they can be called. Run:
ToolSearch query="select:mcp__claude_ai_Linear__save_issue,mcp__claude_ai_Linear__get_issue"
Calling either tool without loading its schema first will fail with
InputValidationError.
The Linear MCP server requires real newlines in markdown content, not
literal \n escape sequences. Pass description strings with actual
line breaks.
If the new issue is a follow-up, cleanup, child, or otherwise tied to
an existing issue, call mcp__claude_ai_Linear__get_issue on that
issue first and reuse:
team IDproject IDprojectMilestone ID (if set)Don't ask the user for fields you can pull from the related issue.
When the user references issues for parent, blocks, or related,
use the Linear identifier format (e.g. DRWFA-1234), not the internal
UUID.
titleteam (UUID)Everything else is optional. Set what you can infer; ask about the rest only when it matters.
Use AskUserQuestion for fields that can't be inferred and are likely
to matter. Typical asks:
Skip the question when context makes the answer obvious. Don't ask about the team or project if you inherited them.
## headings (e.g. ## What to remove,
## What to keep, ## When, ## Context).## Context
section so the reader can find the source of truth.@displayName to mention people.get_issue on the related issue and capture team / project /
milestone IDs.mcp__claude_ai_Linear__save_issue with the collected fields.DRWFA-1234 — <title> (assignee: X, priority: Y, due: Z).Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
Spin up throwaway service/database stacks for tests, migrations, and ingest without touching the user's real dev stack or live data. Use when a task runs `docker compose`, database migrations, ingest jobs, or any destructive command that could hit a shared or live service. Triggers on "run the migrations", "ingest the data", "spin up the stack", "run the integration tests", or any plan step that starts containers or writes to a database.
Run the `pi` coding agent as a headless subagent (--mode json) inside a visible tmux pane so the user can watch it stream while the agent still captures and verifies its structured output. Use when asked to "run pi in a pane", "let me watch pi work", "run pi where I can see it", or to delegate to pi with a live view. Requires running inside tmux. For a fully headless run with no pane, use use-pi-subagent instead.
Invoke the `pi` coding agent as a non-interactive subagent from the shell — delegate a self-contained coding or analysis task to a fresh pi process and read its structured JSON output. Use when asked to "run pi", "use pi as a subagent", "delegate this to pi", or to orchestrate pi coder/reviewer agents (see prompt.md).
Keep the computer awake for a set duration or while a command runs, using `caffeinate` on macOS. Use when the user wants to prevent sleep during a long-running task, download, or presentation.
Writes a self-contained handoff document that briefs another agent or person to execute a scoped task without the originating context. Triggers when the user asks to "write a handoff", "hand this off", "brief another agent", "create a delegation doc", or prep work for a sub-agent / worktree agent to pick up. Emits a markdown file following a fixed template.