claude-prompt-engineering

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Claude Prompt Engineering

Claude提示词工程

Knowledge snapshot from: 2026-02-20

Generated by: cogworks

知识快照时间： 2026-02-20

生成者： cogworks

Overview

概述

This skill provides practical, Claude-specific prompt engineering guidance for Opus 4.6, Sonnet 4.5, and Haiku 4.5. It emphasizes fast model-aware decisions: reasoning mode selection, context management, safe autonomy, tool efficiency, and output control.

本技能为Opus 4.6、Sonnet 4.5和Haiku 4.5提供实用的、针对Claude的提示词工程指导。它强调基于模型感知的快速决策：推理模式选择、上下文管理、安全自主性、工具效率以及输出控制。

When to Use This Skill

适用场景

Use this skill when you need to:

Design or review Claude system prompts
Tune adaptive or extended thinking behavior
Improve tool orchestration and parallelization
Handle long-horizon or multi-window workflows
Add prompt-injection and data-leakage safeguards
Reduce verbosity while preserving answer quality

当你需要以下操作时，可使用本技能：

设计或审核Claude系统提示词
调整自适应或扩展思维行为
优化工具编排与并行处理
处理长周期或多窗口工作流
添加提示词注入和数据泄露防护措施
在保证回答质量的同时降低冗余度

Quick Decision Cheatsheet

快速决策速查表

Opus 4.6: Use adaptive thinking (
```
low|medium|high|max
```
) only when task complexity justifies it.
Sonnet 4.5: Use extended thinking for legacy workflows; minimum budget 1024 tokens.
Simple tasks: Prefer no explicit thinking config.
Long tasks: Persist state in files + git checkpoints.
Independent reads/searches: Run tool calls in parallel.
Risky/irreversible actions: Ask for confirmation first.
Production exposure: Apply defense-in-depth (input, architecture, output).

Opus 4.6：仅当任务复杂度需要时，使用自适应思维（
```
low|medium|high|max
```
）。
Sonnet 4.5：针对遗留工作流使用扩展思维；最低预算为1024 tokens。
简单任务：建议不使用明确的思维配置。
长任务：在文件中保存状态并通过git checkpoint记录。
独立读取/搜索：并行运行工具调用。
高风险/不可逆操作：先请求确认。
生产环境部署：应用纵深防御（输入、架构、输出层面）。

Supporting Docs

配套文档

```
reference.md
```
: Canonical guidance, decision rules, anti-patterns, and sources
```
patterns.md
```
: Reusable patterns and templates
```
examples.md
```
: Compact before/after prompt examples

```
reference.md
```
：标准指导、决策规则、反模式及来源
```
patterns.md
```
：可复用的模式与模板
```
examples.md
```
：简洁的提示词优化前后示例

Model Routing Contract

模型路由约定

primary-capability-class: reasoning
fallback-capability-class: workhorse
task-to-capability mapping:
- source ingestion/extraction: workhorse
- synthesis/contradiction resolution: reasoning
- final skill drafting: workhorse
quality gates tied to capability:
- reasoning: resolve contradictions and justify the interpretation
- workhorse: complete structure with citations and no stubs
runtime model resolution:
- map capability classes to provider/runtime defaults automatically
- never ask the user to choose a model

核心能力类别：推理
** fallback能力类别**：通用处理
任务与能力映射：
- 源数据摄入/提取：通用处理
- 合成/矛盾解决：推理
- 最终技能起草：通用处理
与能力绑定的质量门槛：
- 推理：解决矛盾并解释推理依据
- 通用处理：完成结构化内容并包含引用，无残缺内容
运行时模型解析：
- 自动将能力类别映射到供应商/运行时默认值
- 绝不要求用户选择模型

Invocation

调用方式

text

/claude-prompt-engineering

text

/claude-prompt-engineering