配置选项概述#
Skills 提供了丰富的配置选项，允许用户根据具体需求定制 Skills 的行为。本节将详细介绍各种配置选项及其使用方法。
配置层级#
1. 全局配置#
1.1 配置文件位置
全局配置文件#
文件位置#
- Linux/macOS: ~/.claude/config.yaml
- Windows: %USERPROFILE%.claude\config.yaml
配置示例#
1.2 全局配置项
2. 项目配置#
2.1 配置文件位置
yaml
项目配置项详解
project:
项目元数据
metadata:
name: "my-project"
description: "A web application"
version: "1.0.0"
author: "Team"
技术栈
tech_stack:
language: "python"
version: "3.9"
frameworks:
- name: "flask"
version: "2.3.0"
- name: "sqlalchemy"
version: "2.0.0"
libraries:
- name: "requests"
version: "2.28.0"
- name: "numpy"
version: "1.24.0"
代码规范
code_standards:
style_guide: "PEP8"
formatter: "black"
linter: "pylint"
max_line_length: 88
docstring_style: "google"
测试配置
testing:
framework: "pytest"
coverage_target: 80
test_dir: "tests"
mock_external: true
文档配置
documentation:
format: "markdown"
include_api_docs: true
include_examples: true
output_dir: "docs"
skills:
Skill 覆盖配置
overrides:
code-review:
enabled: true
priority: high
parameters:
strict: true
include_security: true
include_performance: true
3. 用户配置#
3.1 配置文件位置
用户配置文件#
文件位置#
- Linux/macOS: ~/.claude/user.yaml
- Windows: %USERPROFILE%.claude\user.yaml
配置示例#
3.2 用户配置项
用户配置项详解
user:
用户信息
profile:
name: "John Doe"
email: "john@example.com"
timezone: "UTC+8"
language: "zh-CN"
preferences:
交互偏好
interaction:
auto_confirm: false # 自动确认
show_progress: true # 显示进度
show_timing: true # 显示耗时
show_details: false # 显示详细信息
prompt_style: "compact" # 提示样式：compact, verbose
输出偏好
output:
format: "markdown" # 输出格式：markdown, json, html
verbosity: "normal" # 详细程度：quiet, normal, verbose
include_code_blocks: true # 包含代码块
include_line_numbers: false # 包含行号
color_output: true # 彩色输出
max_lines: 1000 # 最大输出行数
编辑器偏好
editor:
default: "vscode" # 默认编辑器
open_in_new_tab: true # 在新标签页打开
focus_after_open: true # 打开后聚焦
line_numbers: true # 显示行号
word_wrap: false # 自动换行
通知偏好
notifications:
enabled: true # 启用通知
on_completion: true # 完成时通知
on_error: true # 错误时通知
sound: false # 播放声音
skills:
Skill 别名
aliases:
cr: code-review
gt: generate-tests
fc: format-code
ref: refactor-code
doc: generate-docs
常用 Skills
favorites:
- code-review
- generate-tests
- format-code
- refactor-code
Skill 默认参数
defaults:
code-review:
strict: false
include_security: true
generate-tests:
framework: pytest
coverage_target: 80
format-code:
style: black
line_length: 88
yaml
code-review Skill 配置
skills:
code-review:
# 基本设置
enabled: true
priority: high
4.2 测试生成配置
generate-tests Skill 配置
skills:
generate-tests:
基本设置
enabled: true
priority: medium
测试框架
framework: "pytest" # pytest, unittest, jest, mocha
测试类型
test_types:
- unit_tests
- integration_tests
- edge_cases
覆盖率目标
coverage:
enabled: true
target: 80
minimum: 60
测试生成策略
generation:
strategy: "comprehensive" # basic, comprehensive, exhaustive
include_positive_tests: true
include_negative_tests: true
include_edge_cases: true
include_boundary_tests: true
Mock 配置
mocking:
enabled: true
mock_external_calls: true
mock_database: true
mock_network: true
use_realistic_mocks: true
测试数据
test_data:
generate_fixtures: true
use_realistic_data: true
include_edge_cases: true
输出配置
output:
test_dir: "tests"
file_naming: "test_{module}.py"
include_docstrings: true
include_comments: true
断言配置
assertions:
use_strict_assertions: true
include_error_messages: true
check_return_values: true
check_exceptions: true
yaml
format-code Skill 配置
skills:
format-code:
# 基本设置
enabled: true
priority: low
配置优先级#
优先级规则#
配置优先级（从高到低）#
1. 命令行参数#
bash
export CLAUDE_SKILL_CODE_REVIEW_STRICT=true
3. 项目配置#
strict: true
~~~
yaml
# ~/.claude/user.yaml
skills:
code-review:
strict: false
### 5. 全局配置
~~~`yaml
````yaml
# ~/.claude/config.yaml
skills:
code-review:
strict: false
```### 6. 默认配置
~~~
yaml
# 内置默认值
strict: false
### 配置合并
~~~
python
class ConfigManager:
def __init__(self):
self.configs = {
"default": self.load_default_config(),
"global": self.load_global_config(),
"user": self.load_user_config(),
"project": self.load_project_config()
}
def get_config(self, skill_name, key):
# 按优先级查找配置
for config_type in ["project", "user", "global", "default"]:
config = self.configs[config_type]
if skill_name in config.get("skills", {}):
skill_config = config["skills"][skill_name]
if key in skill_config:
return skill_config[key]
return None
def merge_configs(self, skill_name):
# 合并所有配置
merged = {}
for config_type in ["default", "global", "user", "project"]:
config = self.configs[config_type]
if skill_name in config.get("skills", {}):
merged = self.deep_merge(merged, config["skills"][skill_name])
return merged
def deep_merge(self, dict1, dict2):
result = dict1.copy()
for key, value in dict2.items():
if key in result and isinstance(result[key], dict) and isinstance(value, dict):
result[key] = self.deep_merge(result[key], value)
else:
result[key] = value
return result
## 动态配置
### 1. 环境变量
## 环境变量配置
### 命名规则
- 前缀: CLAUDE_SKILL_
- 格式: CLAUDE_SKILL_{SKILL_NAME}_{KEY}
- 示例: CLAUDE_SKILL_CODE_REVIEW_STRICT
### 使用示例
~~~`bash
`bash
# 设置代码审查严格模式
export CLAUDE_SKILL_CODE_REVIEW_STRICT=true
# 设置测试生成覆盖率目标
export CLAUDE_SKILL_GENERATE_TESTS_COVERAGE_TARGET=90
# 设置格式化工具
export CLAUDE_SKILL_FORMAT_CODE_FORMATTER=black
```> >
~~~
### 2. 运行时配置
class RuntimeConfig:
def __init__(self):
self.config = {}
def set(self, key, value):
self.config[key] = value
def get(self, key, default=None):
return self.config.get(key, default)
def update(self, updates):
self.config.update(updates)
def clear(self):
self.config.clear()
# 使用示例
runtime_config = RuntimeConfig()
runtime_config.set("code_review.strict", True)
runtime_config.set("generate_tests.coverage_target", 90)
~~~
### 3. 条件配置
~~~
yaml
# 条件配置示例
skills:
code-review:
# 根据环境配置
environments:
development:
strict: false
include_performance: false
testing:
strict: true
include_performance: true
production:
strict: true
include_performance: true
include_security: true
# 根据文件类型配置
file_types:
python:
check_type_hints: true
check_docstrings: true
javascript:
check_eslint: true
check_typescript: true
html:
check_accessibility: true
check_seo: true
## 配置验证
### 1. 配置验证
class ConfigValidator:
def __init__(self):
self.schemas = self.load_schemas()
def validate(self, skill_name, config):
schema = self.schemas.get(skill_name)
if not schema:
return True
errors = []
for key, value in config.items():
if key not in schema:
errors.append(f"Unknown configuration key: {key}")
continue
key_schema = schema[key]
if not self.validate_value(value, key_schema):
errors.append(f"Invalid value for {key}: {value}")
return len(errors) == 0, errors
def validate_value(self, value, schema):
value_type = schema.get("type")
if value_type == "boolean":
return isinstance(value, bool)
elif value_type == "integer":
return isinstance(value, int)
elif value_type == "string":
return isinstance(value, str)
elif value_type == "enum":
return value in schema.get("values", [])
return True
def load_schemas(self):
return {
"code-review": {
"strict": {"type": "boolean"},
"include_security": {"type": "boolean"},
"include_performance": {"type": "boolean"},
"strictness": {"type": "enum", "values": ["low", "medium", "high", "strict"]}
},
"generate-tests": {
"framework": {"type": "enum", "values": ["pytest", "unittest", "jest", "mocha"]},
"coverage_target": {"type": "integer"},
"mock_external_calls": {"type": "boolean"}
}
}
~~~
### 2. 配置迁移
~~~
python
class ConfigMigrator:
def __init__(self):
self.migrations = {
"1.0": self.migrate_1_0_to_1_1,
"1.1": self.migrate_1_1_to_1_2
}
def migrate(self, config, from_version, to_version):
current_version = from_version
while current_version != to_version:
if current_version in self.migrations:
config = self.migrations[current_version](config)
current_version = self.increment_version(current_version)
else:
break
return config
def migrate_1_0_to_1_1(self, config):
# 迁移逻辑
if "skills" in config:
for skill_name, skill_config in config["skills"].items():
if "strict" in skill_config:
skill_config["strictness"] = "high" if skill_config["strict"] else "low"
return config
def increment_version(self, version):
major, minor = version.split(".")
return f"{major}.{int(minor) + 1}"
Skills 的配置选项提供了灵活的定制能力，通过合理的配置可以显著提高 Skills 的适用性和效率。理解配置层级、优先级和验证机制有助于更好地管理和使用 Skills。
在下一节中，我们将探讨 Skills 的生命周期管理，了解 Skills 的创建、更新、删除等操作。