AgentSkillsCN

linear-plan

当仅需计划执行而无需编写代码时，使用此技能。

SKILL.md
--- frontmatter
name: linear-plan
description: Use this skill when planning-only execution is requested and issue context should be created or updated without coding.
version: 1.0.0

Issue Planning Phase

Use this skill for planner-only sessions.

Trigger intent (simple UX)

  • •linear-plan <feature request>
  • •linear-plan <problem statement>

If the request is ambiguous, ask clarification questions first, then continue planning.

Required workflow

  1. •Ask targeted clarification questions until objective and definition of done are clear.
  2. •Define in-scope and out-of-scope boundaries.
  3. •Identify constraints and invariants from docs/agent/security.md and docs/agent/quality.md.
  4. •Define acceptance checks.
  5. •Decompose by ownership (client/server/shared/reviewer/verifier).
  6. •Choose ticket strategy:
    • •small/bounded: create one implementation issue
    • •medium/large or dependency-heavy: create orchestration parent issue plus child implementation/review/verification issues
  7. •Set execution routing label on the main issue:
    • •execution-mode:single for direct execution
    • •execution-mode:orchestrated for parent-child orchestration
  8. •Set deterministic decision labels on the main issue:
    • •risk:low | risk:medium | risk:high
    • •scope:client | scope:server | scope:shared | scope:fullstack
    • •verification:standard | verification:strict
  9. •Set issue hierarchy label on the main issue:
    • •issue-role:top-level (required for Linear view filtering)
  10. •Add done-when checklist in the main issue body (3-6 concrete acceptance bullets).
  11. •Create or update issue(s) with objective, scope, checklist, and dependencies.
  12. •Only for exceptional high-risk/audit work, also create a companion exec-plan file under docs/agent/exec-plans/active/.

Output contract

Always return:

  • •issue references (DIM-123 and URL)
  • •main issue ID for execution
  • •execution routing label set on main issue (execution-mode:single or execution-mode:orchestrated)
  • •decision labels set on main issue (risk:*, scope:*, verification:*)
  • •issue hierarchy label set on main issue (issue-role:top-level)
  • •done-when checklist added to main issue body
  • •selected ticket strategy (single issue vs orchestration parent + children)
  • •concise scope summary
  • •execution checklist for implementer
  • •explicit blockers/open questions
  • •next command hint for user: linear-implement <MAIN-ISSUE-ID>