prose-synthesize

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/prose:synthesize

Synthesize unstructured thinking into a structured, actionable plan — impose order on chaos.

将无结构化的思考转化为结构化、可执行的计划——为混乱的内容赋予秩序。

When to Use This Skill

何时使用此技能

Use this skill when...	Use something else when...
User dumps stream-of-consciousness thoughts	Text is already structured but verbose → `/prose:distill`
Scattered notes need organizing into a plan	Need to capture session learnings → `/project:distill`
Brain dump needs goals, actions, priorities extracted	Need to write a plan from scratch with no input → `/blueprint:init`
User says "make sense of this", "organize this"	Document needs style/tone adjustment → `prose-tone` (planned)

适合使用此技能的场景...	适合使用其他技能的场景...
用户提供意识流式的想法	文本已结构化但过于冗长 → 使用 `/prose:distill`
零散笔记需要整理成计划	需要记录会话中的学习内容 → 使用 `/project:distill`
头脑风暴内容需要提取目标、行动和优先级	需要从零开始撰写计划且无输入内容 → 使用 `/blueprint:init`
用户说“make sense of this”、“organize this”等短语	需要调整文档的风格/语气 → 使用 `prose-tone` （规划中）

Core Principles

核心原则

Synthesis is the complement of analysis. Analysis breaks apart; synthesis combines fragments into a coherent whole. The input is scattered thinking — the output is structured intent.

整合是分析的互补过程。分析是拆解内容；整合是将碎片组合成连贯的整体。输入是零散的思考内容，输出是结构化的行动计划。

What Makes a Good Synthesized Plan

优质整合计划的标准

Each element answers a specific question:

Element	Question it answers
Objective	What are we trying to achieve?
Key decisions	What needs deciding before acting?
Actions	What specific things need doing?
Dependencies	What blocks what?
Open questions	What's still unclear?

每个组成部分都要回答特定问题：

组成部分	回答的问题
目标	我们想要实现什么？
关键决策	行动前需要确定哪些事项？
行动	需要具体做哪些事情？
依赖关系	哪些事项会阻碍其他事项？
待解决问题	哪些内容仍不明确？

Parameters

参数解析

Parse

$ARGUMENTS

If prose is provided inline, synthesize it directly
If a file path is provided, read and synthesize the file contents
If no arguments, ask the user to provide their unstructured thinking

解析

$ARGUMENTS

：

如果直接提供了文本内容，直接对其进行整合
如果提供了文件路径，读取并整合文件内容
如果没有提供参数，请用户提供其无结构化的思考内容

Execution

执行流程

Execute this synthesis workflow:

执行以下整合工作流程：

Step 1: Absorb the input

步骤1：吸收输入内容

Read all provided text without editing or filtering. Identify:

Total volume (rough word count)
Density of ideas — scattered vs. loosely organized
Whether there's an implicit goal or the user is still exploring
Tone — brainstorming, planning, venting, or exploring

阅读所有提供的文本，不进行编辑或过滤。识别：

总篇幅（大致字数）
想法的密集程度——零散型还是松散组织型
是否存在隐含目标，或者用户仍在探索阶段
语气——头脑风暴、规划、宣泄还是探索

Step 2: Extract and tag ideas

步骤2：提取并标记想法

Work through the input and internally classify each distinct idea:

GOAL — something the user wants to achieve
CONSTRAINT — a limitation or boundary condition
DECISION — a choice that needs making
ACTION — a concrete step that could be taken
QUESTION — something unresolved or uncertain
CONTEXT — background that informs the plan
ASIDE — tangential thought (preserve but deprioritize)

Do not discard anything yet. Do not show tags to the user.

梳理输入内容，对每个独特的想法进行内部分类：

GOAL（目标）——用户想要实现的事项
CONSTRAINT（约束）——限制条件或边界
DECISION（决策）——需要做出的选择
ACTION（行动）——可以采取的具体步骤
QUESTION（问题）——未解决或不确定的事项
CONTEXT（背景）——为计划提供信息的背景内容
ASIDE（题外话）——无关紧要的想法（保留但降低优先级）

暂时不要丢弃任何内容，也不要向用户展示这些标签。

Step 3: Cluster by theme

步骤3：按主题聚类

Group ideas into natural clusters:

Ideas addressing the same goal
Actions depending on the same decision
Questions blocking the same cluster of actions

Name each cluster with a short descriptive label.

将想法自然地分组为集群：

针对同一目标的想法
依赖同一决策的行动
阻碍同一行动集群的问题

为每个集群添加简短的描述性标签。

Step 4: Find the spine

步骤4：找到核心主线

Identify the core through-line — the primary objective connecting the most clusters. This becomes the plan's backbone. Other clusters are supporting or parallel tracks.

确定核心主线——连接大多数集群的主要目标。这将成为计划的主干。其他集群作为支持或并行分支。

Step 5: Build the plan

步骤5：构建计划

Structure the output using this format. Include only sections that have content:

undefined

使用以下格式构建输出内容，仅包含有内容的部分：

undefined

Objective

目标

<1-2 sentence statement of what this plan achieves>

<1-2句话说明此计划要实现的内容>

Key Decisions

关键决策

<执行前或执行过程中需要做出的决策。如果用户提到了选项，请注明。>

Plan

计划

<按主题/阶段分组的有序行动。序列使用编号列表，并行事项使用项目符号。>

Dependencies

依赖关系

<哪些事项会阻碍其他事项——仅当存在有意义的依赖关系时才包含>

Open Questions

待解决问题

<计划完成前需要解答的未解决事项>

Parked Ideas

暂存想法

<Tangential thoughts worth preserving but outside the core plan> ```

<值得保留但不属于核心计划的无关想法>

undefined

Step 6: Validate completeness

步骤6：验证完整性

Check the plan against the original input:

Is every goal addressed?
Is every constraint respected?
Are the user's questions surfaced, not buried?
Did anything important get lost in synthesis?

If the input was exploratory (no clear goal), say so in the objective and frame the plan as "potential directions" rather than commitments.

对照原始输入内容检查计划：

是否所有目标都已覆盖？
是否所有约束都已遵守？
用户的问题是否都已明确提出，而非被隐藏？
是否有重要内容在整合过程中丢失？

如果输入内容是探索性的（无明确目标），请在目标部分说明，并将计划表述为“潜在方向”而非承诺。

Handling Ambiguity

处理模糊性

Situation	Approach
Multiple possible goals	Present as alternatives under Key Decisions
Contradictory ideas	Surface the contradiction explicitly
Vague but directional	Interpret the direction, note the interpretation
Pure exploration (no goal)	Organize by theme, suggest possible goals

场景	处理方法
存在多个可能的目标	在“关键决策”部分作为备选方案呈现
存在矛盾的想法	明确指出矛盾之处
模糊但有方向性	解读方向，并注明解读内容
纯探索性内容（无目标）	按主题组织，提出可能的目标

Example

示例

Input: "I need to fix the auth system. Also the tests are broken. Maybe we should move to JWT. The deployment pipeline keeps failing too. Sarah mentioned something about rate limiting. We should probably do a security audit at some point. Oh and the docs are out of date."

Synthesized:

Objective: Stabilize and secure the authentication system while addressing related infrastructure issues.

Key Decisions:

JWT migration — move to JWT or fix current auth? (Affects scope of all auth work)

Plan:

Fix immediate blockers

Fix broken tests (unblocks everything else)

Fix deployment pipeline failures

Auth system

Decide JWT migration

Implement fix or migration based on decision

Add rate limiting (per Sarah's input)

Hardening

Security audit

Update documentation

Open Questions:

What specifically did Sarah say about rate limiting?

What's breaking in the deployment pipeline? (may be auth-related)

输入： "I need to fix the auth system. Also the tests are broken. Maybe we should move to JWT. The deployment pipeline keeps failing too. Sarah mentioned something about rate limiting. We should probably do a security audit at some point. Oh and the docs are out of date."

整合后的结果：

目标： 稳定并加固认证系统的安全性，同时解决相关的基础设施问题。

关键决策：

JWT迁移——是迁移到JWT还是修复当前的认证系统？（影响所有认证工作的范围）

计划：

解决即时障碍

修复损坏的测试（解除所有其他工作的阻塞）

修复部署流水线的故障

认证系统优化

决定是否进行JWT迁移

根据决策实施修复或迁移

添加速率限制（根据Sarah的建议）

系统加固

进行安全审计

更新文档

待解决问题：

Sarah关于速率限制的具体建议是什么？

部署流水线故障的具体原因是什么？（可能与认证相关）

Agentic Optimizations

Agent优化

Context	Approach
Short input (< 200 words)	Synthesize inline, present plan directly
Medium input (200-1000 words)	Full synthesis workflow, structured plan output
Long input or file (> 1000 words)	Read file, full synthesis, write plan to file
Already semi-structured input	Preserve existing structure, fill gaps
No clear goal in input	Organize by theme, present as exploration map

场景	处理方法
短输入（<200词）	直接在线整合，直接呈现计划
中等长度输入（200-1000词）	执行完整的整合工作流程，输出结构化计划
长输入或文件（>1000词）	读取文件，执行完整整合，将计划写入文件
已半结构化的输入	保留现有结构，填补空白
输入无明确目标	按主题组织，呈现为探索地图