plan-mode

GitHub

用于非平凡任务实施前的系统性规划。通过探索、澄清、设计、审查等步骤,确保变更符合架构一致性。适用于新功能、架构决策或多文件变更场景。

.agents/skills/plan-mode/SKILL.md nolly-studio/eve-directory

触发场景

/plan plan this design an approach let's plan first

安装

npx skills add nolly-studio/eve-directory --skill plan-mode -g -y
更多选项

非标准路径

npx skills add https://github.com/nolly-studio/eve-directory/tree/main/.agents/skills/plan-mode -g -y

不安装直接使用

npx skills use nolly-studio/eve-directory@plan-mode

指定 Agent (Claude Code)

npx skills add nolly-studio/eve-directory --skill plan-mode -a claude-code -g -y

安装 repo 全部 skill

npx skills add nolly-studio/eve-directory --all -g -y

预览 repo 内 skill

npx skills add nolly-studio/eve-directory --list

SKILL.md

Frontmatter
{
    "name": "plan-mode",
    "description": "Holistic, system-aware planning before implementing non-trivial tasks. Use when the task involves new features, architectural decisions, multi-file changes, unclear requirements, or multiple valid approaches. Triggers on \"\/plan\", \"plan this\", \"design an approach\", \"let's plan first\"."
}

For non-trivial implementation tasks, plan before you code. Treat every change as a system change, not an isolated patch. Find the smallest implementation that remains coherent with the architecture, lifecycle, and future evolution of the system.

Planning Doctrine

  • Think broadly, implement narrowly. Analyze the surrounding system before deciding where the change belongs.
  • Every change has system implications. Consider data flow, ownership of state, interfaces, lifecycle, failure modes, and future extension.
  • Prefer authoritative, derivable designs. If state can be derived from persisted state or existing abstractions, prefer that over inventing transient duplicates.
  • Choose the smallest coherent solution. Do not default to the smallest local patch if it distorts the architecture.
  • Expand scope only when it simplifies the system. Not for speculative improvement.
  • Keep the language current. When a plan coins or clarifies a domain term, or settles a directional decision, update CONTEXT.md in the same change.

Steps

  1. Explore. Read relevant files; trace the end-to-end flow; identify the current source of truth, state ownership, and lifecycle (trigger, processing, completion, failure, retry, cleanup). Search for prior art. Do not implement yet.
  2. Clarify. Ask the user about ambiguities — behavior, edge cases, failure handling, tradeoffs — before committing to an approach.
  3. Design. Compare the most direct implementation against the most system-coherent one. Answer explicitly: What is the source of truth before/after? What new state or invariants does this introduce? Does it duplicate logic anywhere? What is the smallest solution that keeps the system coherent?
  4. Review. Re-read the critical files your design depends on to verify assumptions. Confirm the design covers loading, in-progress, completion, failure, retry, and cleanup.
  5. Present. Write the plan (structure per docs/plans/README.md) to PLAN.md or docs/plans/<feature>.md and wait for approval.
  6. Implement and verify. Track progress per item, then run the plan's verification steps.

版本历史

  • 80d83bc 当前 2026-07-31 13:42

同 Skill 集合

.agents/skills/bootstrap-agents-md/SKILL.md
.agents/skills/code-review/SKILL.md
.agents/skills/frontend-design/SKILL.md
.agents/skills/setup/SKILL.md
.claude/skills/integration-nextjs-app-router/SKILL.md
.agents/skills/design-md/SKILL.md
.agents/skills/frontier-interview/SKILL.md

元信息

文件数
0
版本
80d83bc
Hash
7293e472
收录时间
2026-07-31 13:42

首页 - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-08-01 00:34
浙ICP备14020137号-1 $访客地图$