
dogkeeper886/testlink-mcp
Developer toolsこのページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る
A Model Context Protocol (MCP) server for TestLink
概要
Create suites, write and refine test cases, build test plans, record executions, and link requirements, all through natural-language chat with any Model Context Protocol client. No more clicking through the TestLink UI. TestLink is powerful but slow to drive by hand — every case edit, plan, and execution is a trip through the web UI. testlink-mcp puts a clean in front of TestLink's XML-RPC API so an AI assistant can do that work for you: - "Add detailed steps to MFT-12 and mark it high importance." - — Claude Code, Cursor, Dify, or your own. - — cases, suites, plans, builds, executions, requirements, projects. - — strict input validation, clear XML-RPC error mapping, and correct internal-vs-external ID handling so edits land on the right case. - docker run and you're connected. You ask in plain language.
README
以下は原文 README の抜粋です。原文を見る。
TestLink MCP Server
Run your TestLink test management from Claude — in plain English. Create suites, write and refine test cases, build test plans, record executions, and link requirements, all through natural-language chat with any Model Context Protocol client. No more clicking through the TestLink UI.
Why testlink-mcp?
TestLink is powerful but slow to drive by hand — every case edit, plan, and execution is a trip through the web UI. testlink-mcp puts a clean MCP server in front of TestLink’s XML-RPC API so an AI assistant can do that work for you:
- Talk, don’t click. “Add detailed steps to MFT-12 and mark it high importance.”
- Works with any MCP client — Claude Code, Cursor, Dify, or your own.
- 30 tools across the whole TestLink model — cases, suites, plans, builds, executions, requirements, projects.
- Solid where it counts — strict input validation, clear XML-RPC error mapping, and correct internal-vs-external ID handling so edits land on the right case.
- One container to run it.
docker runand you’re connected.
How it works
You ask in plain language. Your MCP client (Claude) picks the right tool and arguments; testlink-mcp validates them and makes the matching XML-RPC call to TestLink; the result comes back as a plain-language answer. The server speaks MCP over stdio and TestLink over XML-RPC — you only ever see the conversation.
Quick Start
The published Docker image is the fastest way in.
Claude Code
claude mcp add testlink -- docker run --rm -i \
-e TESTLINK_URL=http://your-testlink-host/testlink \
-e TESTLINK_API_KEY=your_api_key_here \
dogkeeper886/testlink-mcp:latest
Cursor / other MCP clients — add to your MCP config:
{
"mcpServers": {
"testlink": {
"command": "docker",
"args": ["run", "--rm", "-i",
"-e", "TESTLINK_URL=http://your-testlink-host/testlink",
"-e", "TESTLINK_API_KEY=your_api_key_here",
"dogkeeper886/testlink-mcp:latest"]
}
}
}
Then ask your assistant: “List the test suites in project 1.”
Get your API key: in TestLink, My Settings → API interface → Generate key, and make sure the XML-RPC API is enabled for your installation.
What you can manage
testlink-mcp covers the full TestLink object model — the requirement that covers a case, the case that lives in a suite and a plan, the build that belongs to a plan, and the execution that records a result.
インストール
docker run --rm -i -e TESTLINK_URL=http://your-testlink-host/testlink -e TESTLINK_API_KEY=your_api_key_here dogkeeper886/testlink-mcp:latest設定
{
"mcpServers": {
"testlink": {
"command": "docker",
"args": ["run", "--rm", "-i",
"-e", "TESTLINK_URL=http://your-testlink-host/testlink",
"-e", "TESTLINK_API_KEY=your_api_key_here",
"dogkeeper886/testlink-mcp:latest"]
}
}
}