Qase MCP Server

Official Qase MCP server. Manage test cases, runs, plans, suites, results, defects, and milestones through AI assistants with full QQL search support.

Developer Tools by Qase TMS API Key active

GitHub Docs npm: @qase/mcp-server

Overview

The Qase MCP Server is the official Model Context Protocol integration for the Qase test management platform. It connects AI assistants like Claude Desktop, Cursor, Claude Code, OpenAI Codex, and OpenCode to your Qase workspace, exposing the full Qase API through a standardized protocol so agents can read, create, and update test artifacts without leaving the IDE or chat surface.

The server ships with roughly 83 tools spanning every major Qase entity: projects, test cases, test suites, test runs, test results, test plans, defects, milestones, environments, shared steps and parameters, attachments, custom fields, configurations, and users. It includes first-class support for QQL (Qase Query Language) so agents can filter cases, defects, and results by arbitrary fields, and supports enterprise installations with custom API domains and custom CA certificates.

Tests created through the MCP server are stored in Qase like any other artifact, inheriting suites, runs, role-based access, audit trails, milestones, and approval workflows. This makes it suitable for bringing AI-generated tests into a real SDLC rather than keeping them in a parallel sandbox. The server is implemented in TypeScript and supports stdio, SSE, and HTTP transports.

Tools

Tool	Description
`Projects (6 tools)`	List, get, create, delete projects and grant or revoke project access.
`Test Cases (8 tools)`	List, get, create, update, delete test cases, bulk create, and link issues.
`Test Runs (7 tools)`	List, get, create, delete, and complete test runs and manage public links.
`Test Results (6 tools)`	List, get, create, update, delete, and bulk create test results.
`Test Plans (5 tools)`	CRUD operations for test plans.
`Test Suites (5 tools)`	CRUD operations for test suites used to organize cases.
`Defects (7 tools)`	List, get, create, update, delete, resolve, and update statuses of defects.
`Milestones (5 tools)`	CRUD operations for milestones.
`Environments (5 tools)`	CRUD operations for environments.
`Shared Steps and Parameters (10 tools)`	CRUD operations for shared steps and shared parameters reused across cases.
`Attachments (4 tools)`	Upload, list, get, and delete file attachments.
`Custom Fields and System Fields (6 tools)`	Manage custom field schemas and retrieve system field info.
`Configurations, Authors, Users (7 tools)`	Retrieve configurations, project authors, and users.
`QQL Search (2 tools)`	Execute Qase Query Language searches across entities and retrieve QQL syntax help.

Setup Guide

Prerequisites

Node.js 18 or later
A Qase account and API token from app.qase.io/user/api/token

Install

The recommended approach is to run via npx so you always get the latest published version:

npm install -g @qase/mcp-server

Or run from source:

git clone https://github.com/qase-tms/qase-mcp-server.git
cd qase-mcp-server
npm install && npm run build

Claude Desktop config

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on your OS:

{
  "mcpServers": {
    "qase": {
      "command": "npx",
      "args": ["-y", "@qase/mcp-server"],
      "env": {
        "QASE_API_TOKEN": "your_api_token_here"
      }
    }
  }
}

Claude Code

claude mcp add qase -- npx -y @qase/mcp-server
export QASE_API_TOKEN=your_api_token_here

Cursor

Use the same JSON config as Claude Desktop in Cursor's MCP settings.

Environment variables

QASE_API_TOKEN (required): API token from your Qase account.
QASE_API_DOMAIN (optional): Defaults to api.qase.io. Set to your custom domain for enterprise deployments.
NODE_EXTRA_CA_CERTS (optional): Path to a PEM certificate file when running behind a corporate proxy with custom SSL.

Transports

The server supports stdio (default), SSE (npm run start:sse), and HTTP (npm run start:http) transports on port 3000. Use node build/index.js --transport [type] --port [num] --host [addr] for custom values.

Use Cases

Generate test cases from a feature spec or pull request diff and create them in the right Qase suite without leaving Cursor or Claude Code.
Trigger a test run for a milestone, post manual results from a debugging session, and complete the run once acceptance is signed off.
Query failing or flaky tests with QQL (for example, status=failed AND author=me AND updated_at>=last_week) and summarize patterns for a standup.
File and triage defects directly from agent conversations, linking them back to the test cases that surfaced the regression.
Bulk import test cases produced by an AI agent into a new project, including suites, shared steps, and custom fields.

Example Prompts

"Create a test plan in project DEMO covering all smoke test cases tagged @critical and start a new run against the staging environment."
"Use QQL to find every test case in project APP that failed in the last 3 runs and group them by suite."
"Draft 10 test cases for the new password reset flow described in this PR and add them to the Auth suite in Qase."
"Open a defect titled 'Checkout 500 on coupon apply', severity major, and link it to case ID 142."
"Show me all open milestones in project WEB and their associated test run completion percentages."

Pros

Official server maintained by Qase TMS with complete coverage of the Qase API across 80+ tools.
First-class QQL support lets agents run precise queries across cases, defects, and results.
Multiple transport options (stdio, SSE, HTTP) and enterprise features like custom API domains and custom CA certificates.
Distributed as an npm package (@qase/mcp-server), so npx based setup works in Claude Desktop, Cursor, Claude Code, Codex, and OpenCode.

Limitations

Requires a Qase account and API token; advanced features and seat limits depend on your Qase plan.
The large surface area (83 tools) can crowd an agent's context window; you may want to scope which tools are exposed per task.
No built-in OAuth flow: authentication relies on long-lived API tokens managed via environment variables.

Alternatives

QA Sphere MCP for the QA Sphere TMS.
rikuson/mcp-qase, a community Qase MCP server that predates the official one.
TestRail and Zephyr have community MCP integrations available on GitHub for teams on those platforms.