zephyr-mcp
Enables reading test cases and test steps from Zephyr Scale using the Zephyr Scale API.
README
Zephyr MCP
An MCP server for Zephyr Scale. The current version provides tools for reading test cases and test steps.
Requirements
- Node.js 20.6 or later
- A Zephyr Scale API token
- Codex CLI or Claude Code CLI
- Bash on Linux or macOS
- Git Bash or WSL on Windows
On Windows, use Git Bash when Node.js and the MCP client are installed on Windows. Use WSL only when Node.js and the Codex or Claude Code CLI are also installed inside the same WSL environment.
1. Get the repository
Clone the repository and enter its directory:
git clone <repository-url>
cd zephyr-mcp
The setup scripts resolve the repository from their own location. After the repository has been cloned, the scripts can be launched from any working directory.
2. Create the repository .env
Get a Zephyr Scale API token
- Sign in to Jira.
- Open the profile menu in the top-right corner.
- Select Zephyr Scale API Access Tokens.
- Create a token, copy it, and store it securely. Do not commit or share it.

Configure .env
Copy .env.example to .env in the root of the cloned zephyr-mcp
repository:
cp .env.example .env
Then replace the placeholder token:
ZEPHYR_API_TOKEN=your-zephyr-api-token
# Optional
# ZEPHYR_CUSTOM_BASE_URL=https://api.zephyrscale.smartbear.com/v2
The repository ignores .env files. Do not commit your API token.
The setup scripts store only the path to this file in the MCP configuration. They do not copy the token into Codex or Claude Code configuration files.
3. Run the setup script
No separate dependency installation, build, or manual MCP configuration is required. The selected script handles those steps automatically.
If zephyr-mcp was previously installed globally from a .tgz package, append
--global to the setup command to use that package instead of building the
repository source.
Codex
From the repository:
bash setup-mcp-codex.sh
From another directory:
bash /absolute/path/to/zephyr-mcp/setup-mcp-codex.sh
The script registers a user-level zephyr MCP server in Codex. If a server
with that name already exists, it is replaced.
Restart Codex and verify the connection:
/mcp
Claude Code
From the repository:
bash setup-mcp-claude.sh
From another directory:
bash /absolute/path/to/zephyr-mcp/setup-mcp-claude.sh
The script registers a user-scoped zephyr MCP server in Claude Code. If a
server with that name already exists, it is replaced.
Restart Claude Code and verify the connection:
/mcp
You can also inspect the registration from the terminal:
claude mcp get zephyr
What the setup scripts do
Both scripts:
- Resolve the repository path from the script location.
- Require Node.js 20.6 or later.
- Check that
.envexists in the repository. - Verify that
ZEPHYR_API_TOKENis not empty without printing it. - Install missing dependencies and build the local server in default
--sourcemode. - Use the globally installed
.tgzpackage when passed--global. - Normalize paths when running through Git Bash on Windows.
- Replace the existing
zephyrMCP registration. - Configure Node to load the repository
.envusing--env-file.
Troubleshooting
Missing .env
Error: Missing .../.env. Create it with ZEPHYR_API_TOKEN before running this script.
Create .env next to setup-mcp-codex.sh and
setup-mcp-claude.sh.
Missing API token
Error: ZEPHYR_API_TOKEN is missing or empty in .../.env.
Add a non-empty ZEPHYR_API_TOKEN to the repository .env.
MCP startup failure
Run the relevant setup script again:
bash setup-mcp-codex.sh
# or
bash setup-mcp-claude.sh
The script validates .env, rebuilds the server in source mode, and replaces
the existing MCP registration. Restart the MCP client and check /mcp after
the script completes.
Development
npm run typecheck
npm run build
Source code is located in src/; compiled output is written to build/.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。