Aws Knowledge Tool

aws知识工具

](https://www.python.org/downloads/)     

用于查询AWS知识MCP服务器的CLI工具-搜索、阅读和发现AWS文档

目录

• 关于
• 特性
• 安装
• 快速开始
• 命令
• 图书馆使用情况
• 发展
• 已知问题
• 资源

关于

aws-knowledge-tool 是一个 CLI优先 通过以下方式提供对AWS文档的编程访问的工具 AWS知识MCP服务器。它的特点是 代理人友好型设计 具有可组合命令、结构化输出和管道支持。

什么是AWS知识MCP服务器？

AWS知识MCP服务器是一个远程MCP（模型上下文协议）服务器，提供对以下内容的访问：

• AWS文档
• AWS博客
• AWS解决方案库
• AWS架构中心
• AWS规范性指南

服务器: https://knowledge-mcp.global.api.aws\ 协议：MCP（基于HTTP的JSON-RPC 2.0）\ 认证：无（公开，价格有限）

为什么CLI优先？

• 代理商友好结构化命令和错误消息使AI代理（克劳德代码等）能够有效地推理和行动
• 可组合：JSON输出和stderr分离允许轻松的管道和集成
• 可重复使用的：命令是技能、自动化和工作流的构建块
• 可靠的：类型安全、经过测试和可预测的行为

特性

• 🔍 搜索 -使用分页搜索AWS文档
• 📖 阅读 -获取AWS文档并将其转换为markdown
• 💡 推荐 -查找相关文档
• 📊 多级冗长 -渐进式日志记录详细信息（-v/-vv/-vvv）
• 🐚 壳牌完井 -bash、zsh和fish的选项卡完成
• 🔒 安全扫描 -自动秘密检测、代码安全和依赖性检查
• 🤖 代理商友好 -结构化JSON输出，清除错误消息
• 🔗 可组合 -管道工作流的Stdin支持
• 📋 多种格式 -JSON和markdown输出
• 🎯 类型安全 -严格的mypy检查，全面的类型提示
• ✅ 测试良好 -带质量检查的Pytest套件

安装

先决条件

• Python 3.14+
• 紫外线 包管理器

全球安装

# Clone and install
git clone https://github.com/dnvriend/aws-knowledge-tool.git
cd aws-knowledge-tool
uv tool install .

# Verify installation
aws-knowledge-tool --version

# Optional: Install shell completion
eval "$(aws-knowledge-tool completion bash)"  # For bash
eval "$(aws-knowledge-tool completion zsh)"   # For zsh

快速开始

# Search AWS documentation
aws-knowledge-tool search "Lambda function URLs"

# Read a documentation page  
aws-knowledge-tool read "https://docs.aws.amazon.com/lambda/latest/dg/welcome.html"

# Get recommendations
aws-knowledge-tool recommend "https://docs.aws.amazon.com/lambda/latest/dg/welcome.html" --type new

# Pipeline composition
aws-knowledge-tool search "Lambda" --json | jq -r '.[0].url' | aws-knowledge-tool read --stdin

命令

search-搜索AWS文档

aws-knowledge-tool search QUERY [OPTIONS]

Options:
  -l, --limit INTEGER    Maximum results (default: 10)
  -o, --offset INTEGER   Skip first N results (default: 0)
  --json                 Output JSON format
  --stdin                Read query from stdin
  -v, --verbose          Enable verbose output (use -v for INFO, -vv for DEBUG, -vvv for TRACE)
  -q, --quiet            Suppress non-essential output

Examples:
  # Basic search
  aws-knowledge-tool search "S3 versioning" --limit 5

  # With INFO logging (-v)
  aws-knowledge-tool search "DynamoDB" -v

  # With DEBUG logging and full tracebacks (-vv)
  aws-knowledge-tool search "Lambda" -vv

  # With TRACE logging (includes library internals) (-vvv)
  aws-knowledge-tool search "RDS" -vvv

  # JSON output for pipelines
  aws-knowledge-tool search "Lambda" --json

  # Read from stdin
  echo "Lambda" | aws-knowledge-tool search --stdin

read-阅读AWS文档

aws-knowledge-tool read URL [OPTIONS]

Options:
  -s, --start-index INT  Starting character index
  -m, --max-length INT   Maximum characters to fetch
  --json                 Output JSON format
  --stdin                Read URL from stdin
  -v, --verbose          Enable verbose output (use -v for INFO, -vv for DEBUG, -vvv for TRACE)
  -q, --quiet            Suppress non-essential output

Examples:
  # Read full document
  aws-knowledge-tool read "https://docs.aws.amazon.com/lambda/latest/dg/welcome.html"

  # Read with pagination
  aws-knowledge-tool read "https://docs.aws.amazon.com/..." --start-index 5000 --max-length 2000

  # Pipeline from search
  aws-knowledge-tool search "Lambda" --json | jq -r '.[0].url' | aws-knowledge-tool read --stdin

  # With DEBUG logging
  aws-knowledge-tool read "https://docs.aws.amazon.com/..." -vv

recommended-获取文档建议

aws-knowledge-tool recommend URL [OPTIONS]

Options:
  -t, --type TYPE        Filter by type (highly_rated, new, similar, journey)
  -l, --limit INTEGER    Max results per category (default: 5)
  -o, --offset INTEGER   Skip first N per category (default: 0)
  --json                 Output JSON format
  --stdin                Read URL from stdin
  -v, --verbose          Enable verbose output (use -v for INFO, -vv for DEBUG, -vvv for TRACE)
  -q, --quiet            Suppress non-essential output

Examples:
  # Get all recommendations
  aws-knowledge-tool recommend "https://docs.aws.amazon.com/lambda/latest/dg/welcome.html"

  # Filter by type (find new features)
  aws-knowledge-tool recommend "https://docs.aws.amazon.com/..." --type new

  # Limit results and JSON output
  aws-knowledge-tool recommend "https://docs.aws.amazon.com/..." --json --limit 3

  # With INFO logging
  aws-knowledge-tool recommend "https://docs.aws.amazon.com/..." -v

多级孔隙度测井

使用渐进式详细标记控制日志记录细节。所有日志输出到stderr，保持stdout数据干净。

日志记录级别

示例

# Quiet mode - only errors and warnings
aws-knowledge-tool search "Lambda"

# INFO - see operations and progress
aws-knowledge-tool search "Lambda" -v
# Output:
# [INFO] Search command started
# [INFO] Searching for: Lambda
# [INFO] Found 10 results

# DEBUG - see detailed information and full tracebacks
aws-knowledge-tool search "Lambda" -vv
# Output:
# [INFO] Search command started
# [INFO] Searching for: Lambda
# [DEBUG] Parameters: limit=10, offset=0
# [DEBUG] Connecting to AWS Knowledge MCP Server...
# [DEBUG] Connected. Searching...
# [DEBUG] Search API call complete
# [INFO] Found 10 results

# TRACE - see library internals (MCP, httpx, httpcore, anyio logs)
aws-knowledge-tool search "Lambda" -vvv
# Output: Same as -vv plus httpx request/response logs

与管道集成

记录到stderr允许干净的管道组合：

# Data to stdout, logs to stderr - perfect for pipelines
aws-knowledge-tool search "Lambda" --json -v | jq -r '.[0].url' | aws-knowledge-tool read --stdin -v

壳牌完井

该工具为bash、zsh和fish shell提供原生shell完成，遵循与kubectl和docker等流行工具相同的模式。

支持的外壳

安装

快速设置（临时）

# Bash - active for current session only
eval "$(aws-knowledge-tool completion bash)"

# Zsh - active for current session only
eval "$(aws-knowledge-tool completion zsh)"

# Fish - active for current session only
aws-knowledge-tool completion fish | source

永久设置（推荐）

# Bash - add to ~/.bashrc
echo 'eval "$(aws-knowledge-tool completion bash)"' >> ~/.bashrc
source ~/.bashrc

# Zsh - add to ~/.zshrc
echo 'eval "$(aws-knowledge-tool completion zsh)"' >> ~/.zshrc
source ~/.zshrc

# Fish - save to completions directory
mkdir -p ~/.config/fish/completions
aws-knowledge-tool completion fish > ~/.config/fish/completions/aws-knowledge-tool.fish

基于文件的安装（性能更好）

为了获得更好的shell启动性能，请为文件生成完成脚本：

# Bash
aws-knowledge-tool completion bash > ~/.aws-knowledge-tool-complete.bash
echo 'source ~/.aws-knowledge-tool-complete.bash' >> ~/.bashrc

# Zsh
aws-knowledge-tool completion zsh > ~/.aws-knowledge-tool-complete.zsh
echo 'source ~/.aws-knowledge-tool-complete.zsh' >> ~/.zshrc

# Fish (automatic loading from completions directory)
mkdir -p ~/.config/fish/completions
aws-knowledge-tool completion fish > ~/.config/fish/completions/aws-knowledge-tool.fish

用法

安装完成后，自动完成：

# Tab completion for commands
aws-knowledge-tool 
# Shows: search read recommend completion

# Tab completion for options
aws-knowledge-tool search --
# Shows: --limit --offset --json --stdin --verbose --quiet --help

# Tab completion for shell types
aws-knowledge-tool completion 
# Shows: bash zsh fish

获取帮助

# View completion installation instructions
aws-knowledge-tool completion --help

图书馆使用情况

使用 aws-knowledge-tool 作为Python库：

import asyncio
from aws_knowledge_tool import MCPClient

async def search_docs():
    async with MCPClient() as client:
        results = await client.search_documentation("Lambda functions", limit=5)
        for result in results:
            print(f"{result['title']}: {result['url']}")

asyncio.run(search_docs())

发展

# Clone repository
git clone https://github.com/dnvriend/aws-knowledge-tool.git
cd aws-knowledge-tool

# Install dependencies
make install

# Run quality checks
make format      # Format with ruff
make lint        # Lint with ruff
make typecheck   # Type check with mypy
make test        # Run pytest
make check       # Run all checks (lint + typecheck + test + security)

# Security scanning
make security-bandit      # Python security linter
make security-pip-audit   # Dependency vulnerability scanner
make security-gitleaks    # Secret and API key detection
make security             # Run all security checks

# Build and install
make pipeline    # format + lint + typecheck + test + security + build + install-global

安全扫描

该项目包括三个轻量级安全工具，可提供80%以上的覆盖率：

gitleaks的先决条件：

# macOS
brew install gitleaks

# Linux
# See: https://github.com/gitleaks/gitleaks#installation

所有安全检查在中自动运行 make check 和 make pipeline.

已知问题

MCP HTTP传输

问题：当前的MCP Python SDK实现使用基于stdio的传输（read_stream/write_stream），而不是HTTP传输。该工具目前使用带有内存流的存根实现进行类型检查。

影响：在实施正确的HTTP传输之前，该工具不会连接到AWS知识MCP服务器。

变通方案：实现需要MCP SDK的HTTP客户端包装器或自定义HTTP到流适配器。

全部:

• 调查MCP Python SDK HTTP传输文档
• 使用anyio流或httpx适配器实现正确的HTTP客户端
• 请参阅：https://github.com/modelcontextprotocol/python-sdk

代码位置: aws_knowledge_tool/core/mcp_client.py:39-63

资源

• AWS知识MCP服务器: https://knowledge-mcp.global.api.aws
• MCP规范: https://modelcontextprotocol.io
• MCP Python SDK: https://github.com/modelcontextprotocol/python-sdk
• AWS文档: https://docs.aws.amazon.com

许可证

MIT许可证-请参阅 许可证 文件

作者

丹尼斯 朋友\ github： @dn朋友

______________________________________________________________________

🤖 内置于 克劳德代码

该项目是使用Anthropic的人工智能开发工具Claude Code开发的。