AgentSkillsCN

create-issues

根据精炼后的项目计划，创建 Linear 问题。确保问题被正确创建，并附带足够的上下文，便于代理顺利执行。可在 /refine-plan 后使用此功能。

SKILL.md
--- frontmatter
name: create-issues
description: >
  Creates Linear issues from a refined project plan. Validates issues
  were created correctly with sufficient context for agent execution.
  Use after /refine-plan.
allowed-tools:
  - mcp__sequential-thinking__sequentialthinking
  - mcp__linear__*
  - Read
  - Write
  - Glob
  - Grep
agent: issue-creator
context: fork

Instructions

You are creating Linear issues from a refined project plan.

Input

  • •The refined plan file path is passed as $ARGUMENTS
  • •If no argument provided, look for docs/project-plan-refined.md

Process

Step 0: Linear Project Setup

Before creating any issues, you must determine the target team and project.

  1. •

    Detect team:

    • •Call list_teams to get all teams in the workspace
    • •If exactly 1 team exists, auto-select it and inform the user: "I'll create issues in team '[name]'."
    • •If multiple teams exist, ask the user to pick one using AskUserQuestion with the team names as options
    • •If 0 teams exist, stop and tell the user to create a team in Linear first
  2. •

    Find or create project:

    • •Extract the project name from the plan file heading (the # [Project Name] at the top of the refined plan)
    • •Search for existing projects in the selected team using list_projects with that name as a query
    • •Exact match found: Ask the user — "I found an existing project called '[name]'. Should I add issues there, or create a new project?"
    • •Multiple partial matches: Show the matches and ask the user to pick one or create a new project
    • •No match found: Tell the user — "I'll create a new Linear project called '[name]'. Want a different name?" Wait for confirmation, then create the project using create_project
  3. •

    Confirm and proceed: Summarize the setup — "Creating issues in team '[team]', project '[project]'." Then continue to issue creation below.

Step 1: Create Issues

  1. •

    Read the refined plan and parse all tasks

  2. •

    For each task, create a Linear issue with:

    • •Team: The team selected in Step 0
    • •Project: The project confirmed/created in Step 0
    • •Title: From the task title
    • •Description: Combine the Context, Steps, and Validation sections into a well-structured issue body. Format it so that an AI agent receiving this issue has everything it needs:
      • •"## Context" — background and architectural decisions
      • •"## Implementation Steps" — the numbered steps
      • •"## Validation Criteria" — how to verify completion
      • •"## Dependencies" — links to dependent issues (add after all issues are created)
    • •Labels: Add complexity label (S/M/L)
    • •Priority: Derive from dependency order (earlier = higher)
  3. •

    After all issues are created:

    • •Go back and update dependency references to use actual Linear issue IDs/links
    • •Set parent/sub-issue relationships where appropriate
  4. •

    Update the refined plan file:

    • •Add the Linear issue ID next to each task title
    • •Add a "Linear Project" link at the top of the document

Output

  • •Linear issues created and linked
  • •Updated docs/project-plan-refined.md with issue IDs
  • •Write a summary to docs/linear-issues-created.md containing:
    • •Total issues created
    • •Issue ID → Task title mapping
    • •Any issues that needed special handling

Validation Criteria

Read .claude/skills/_shared/validation-pattern.md and apply it. Specific criteria for this skill:

  • • Every task in the refined plan has a corresponding Linear issue
  • • Each issue's description contains Context, Implementation Steps, and Validation Criteria sections
  • • Read each issue back from Linear via MCP and verify the content matches what was intended
  • • Dependencies are correctly linked between issues
  • • An agent picking up any single issue would have enough context to: understand the task, complete the work, and validate the result
  • • The refined plan file has been updated with issue IDs