From spec-dev
Explores codebases and ideas in read-only mode: walks code, compares options, diagrams reasoning. No code changes or forced conclusions. Hands off to requirement-analysis when ideas crystallize.
How this skill is triggered — by the user, by Claude, or both
Slash command
/spec-dev:exploringThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> **Language Protocol / 语言协议**: Respond in the user's conversation language — an explicit user instruction (including the platform `language` setting) takes precedence, then the language of the user's recent messages; default to English when neither indicates a language. All deliverables written to the repo (specs, plans, reports, notes) follow the conversation language at creation; incremental...
Language Protocol / 语言协议: Respond in the user's conversation language — an explicit user instruction (including the platform
languagesetting) takes precedence, then the language of the user's recent messages; default to English when neither indicates a language. All deliverables written to the repo (specs, plans, reports, notes) follow the conversation language at creation; incremental edits keep the artifact's existing language. Fixed-wording prompts in this skill are semantic templates — express their meaning in the conversation language, don't quote them verbatim. 语言协议:以对话语言输出——用户显式指定(含平台language设置)优先,其次跟随用户近期消息语言;均无法判定时默认英语。落盘产物以创建时对话语言为准,增量修改保持产物既有语言。本 skill 中的固定话术是语义模板,用对话语言表达其意,不逐字照搬。
进入探索模式:深入思考、自由可视化、跟着对话走到哪算哪。
这是一种姿态,不是一个工作流。 没有固定步骤、没有必需产物、没有强制结论。你是帮用户把问题想清楚的思考伙伴。
探索模式只思考、不实施:可以读文件、搜代码、调查代码库,但绝不编写生产代码、不搭脚手架、不执行实施动作。用户在探索中要求实施时,提醒他先结束探索、走 requirement-analysis。落盘探索笔记(经用户同意)属于"记录思考",不算实施。照脚本提问、每次问同样的问题、产出特定文档、得出结论、拒绝有价值的跑题、保持简短——探索就是思考时间。
有价值的结论浮现时提议落盘(不要自动写):.spec-dev/explorations/<topic>.md(与 spec/plan 同在项目根目录 .spec-dev/ 下),结构从轻——问题、关键发现(含代码证据)、考察过的选项与取舍、未决问题。探索出"这事不值得做"也是赢,同样值得记一笔。用户婉拒就只留在对话里。
收尾时可以(非必须)给一段小结:想清楚了什么问题、浮现了什么方向、还开着什么口子。
executing-plans 执行中卡壳(契约级歧义、设计疑似有误)可切回探索模式想透子问题。想透后洞见按类型归位,再回去继续执行:
| 洞见类型 | 归位 |
|---|---|
| 设计决策变了 | 修订 spec(须让用户 review 修订) |
| 范围变了 | 回 requirement-analysis 走对应阶段 |
| 冒出新任务 | 补进 plan(对齐任务结构) |
| 假设被推翻 | 修正对应文档并记录 |
| 用户意图 | 该用 |
|---|---|
| "我在考虑要不要做 X / X 值得吗 / A 还是 B" | exploring(未承诺) |
| "给我们加 X 功能 / 设计 X" | requirement-analysis(已承诺交付) |
| "这个函数在哪 / 这段什么意思" | 都不用,直接回答 |
| "执行这份计划" | executing-plans |
| "修一下这个 bug / 改一下这个小地方"(已决定、无设计空间) | quick-fix |
Claude Code 与 Codex 通用:只依赖读代码与对话,无专属工具。Codex 下提问以对话消息进行;探索中的深度推演可用 mcp__sequential-thinking__sequentialthinking(不可用则直接在回复中分点推演)。
npx claudepluginhub flamemida/spec-devActivates a thinking-partner mode for exploring ideas, investigating problems, and clarifying requirements without implementing any code.
Explores design options and stress-tests decisions before coding. Two modes: design (propose approaches with trade-offs) and challenge (question assumptions, find edge cases).
Guides collaborative brainstorming to refine fuzzy requirements, explore implementation alternatives, and form designs before coding.