TestRail MCP Server
Connect your AI coding assistant to TestRail to manage test cases, runs, results, and more directly from VS Code, Cursor, Claude Desktop, or Claude Code.
README
TestRail MCP Server
Connect your AI coding assistant to TestRail — manage test cases, runs, results, and more directly from VS Code, Cursor, Claude Desktop, or Claude Code. Built on the Model Context Protocol and powered by testrail_api_module.
Features
- Dynamic tool discovery — endpoints are introspected from
testrail_api_moduleat startup, so new API coverage is picked up automatically - LLM-friendly — instead of registering hundreds of tools, provides a category-based discovery pattern so models can explore available operations without being overwhelmed
Requirements
- Python 3.11+
- uv
Installation
Step 1 — Install tram-mcp once
uv tool install tram-mcp
This puts a tram-mcp binary on your PATH (typically ~/.local/bin/tram-mcp
on macOS/Linux, %APPDATA%\Python\Scripts\tram-mcp.exe on Windows).
tram-mcp then auto-updates itself: on launch (at most once per day) it
spawns a detached uv tool upgrade tram-mcp in the background. The current
process keeps running on the existing binary; the upgrade lands on the next
launch. Set TRAM_MCP_NO_AUTO_UPDATE=1 to disable, or run
uv tool upgrade tram-mcp manually anytime.
Why this and not
uvx tram-mcp?uvxre-resolves and re-extracts wheels on every launch. On Windows this races with Defender / OneDrive scanning the uv cache and intermittently fails withAccess is denied. (os error -2147024891)— the server starts, then exits before it can respond. Even on a healthy machine, a cold uvx cache costs ~4 s+ and can exceed an MCP client's startup timeout. Installing once removes both problems.
Step 2 — Configure your MCP client
Claude Desktop
Add to your Claude Desktop config file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"testrail": {
"command": "tram-mcp",
"env": {
"TESTRAIL_URL": "https://yourinstance.testrail.io",
"TESTRAIL_USERNAME": "your-email@example.com",
"TESTRAIL_API_KEY": "your-api-key"
}
}
}
}
Claude Code
claude mcp add testrail \
-e TESTRAIL_URL=https://yourinstance.testrail.io \
-e TESTRAIL_USERNAME=your-email@example.com \
-e TESTRAIL_API_KEY=your-api-key \
-- tram-mcp
VS Code / VS Code Insiders
Create .vscode/mcp.json in your project (or add to your User Settings):
{
"servers": {
"testrail": {
"type": "stdio",
"command": "tram-mcp",
"env": {
"TESTRAIL_URL": "",
"TESTRAIL_USERNAME": "",
"TESTRAIL_API_KEY": ""
}
}
}
}
VS Code supports ${input:variableName} placeholders to prompt for values at startup.
Cursor
Create .cursor/mcp.json in your project (or ~/.cursor/mcp.json globally):
{
"mcpServers": {
"testrail": {
"command": "tram-mcp",
"env": {
"TESTRAIL_URL": "https://yourinstance.testrail.io",
"TESTRAIL_USERNAME": "your-email@example.com",
"TESTRAIL_API_KEY": "your-api-key"
}
}
}
}
Alternative — no-install via uvx
If you prefer not to maintain a separate install, you can use uvx tram-mcp
as the command (with "args": ["tram-mcp"]) — Cursor / VS Code / etc. style.
This re-resolves on each launch, so a cold cache pays a noticeable ~4 s cost
that can intermittently exceed MCP client startup timeouts. The launch-time
auto-update (see above) also no-ops in this mode — uvx already pulls fresh
on each invocation.
Not recommended on Windows.
uvx's per-launch wheel extraction races Windows Defender real-time scanning and OneDrive sync over the uv cache, producing intermittentFailed to install: <wheel>. Caused by: Access is denied. (os error -2147024891)errors and "server failed to start" behavior that retries fix. Use theuv tool installpath above instead.
Configuration
The server requires TestRail credentials via environment variables:
| Variable | Required | Description |
|---|---|---|
TESTRAIL_URL |
Yes | Your TestRail instance URL (e.g. https://example.testrail.io) |
TESTRAIL_USERNAME |
Yes | TestRail username or email |
TESTRAIL_API_KEY |
Yes* | TestRail API key |
TESTRAIL_PASSWORD |
Yes* | TestRail password (alternative to API key) |
*Either TESTRAIL_API_KEY or TESTRAIL_PASSWORD must be set. API key is recommended.
Development
# Install dependencies
uv sync
# Run the server locally
uv run tram_mcp
# Run tests
uv run pytest
# Run a single test
uv run pytest path/to/test.py::test_name -v
Releasing
Merging a PR to main automatically tags the version and publishes to PyPI. You must bump the version before merging:
uv version 0.6.0 # update version in pyproject.toml
Also bump manifest.json's version to match — both files must agree, and the .mcpb upload is rejected when the version isn't incremented. If you forget to bump, the PyPI publish is skipped (the existing version tag already exists on PyPI).
License
MIT
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。