qai-consultant-mcp — Claude Code MCP server: QA without API keys
A local MCP server for QA work: pulls in standards, estimates effort, reviews documents and analyses test results. Runs without API keys.
claude mcp add qai-consultant-mcp -- uvx qai-consultant-mcpWhat qai-consultant-mcp does
The server hands your agent a QA knowledge base covering ISTQB, OWASP, ISO 26262, A-SPICE, IEEE 829, ISO/IEC 25010 and the EU AI Act. From a short project description the agent can produce a test strategy, a risk register, an effort estimation report and a test plan.
Separate deterministic modes handle what already exists: scoring a test plan or strategy across six ISTQB and IEEE 829 dimensions, parsing JUnit XML or CSV results to surface flaky, ever-failing and slowest tests, and rating process maturity against TMMi. Scores are computed without an LLM call, so the same input gives the same numbers regardless of which model you run.
It starts with uvx qai-consultant-mcp — no API keys, no Pinecone. You need uvx on the machine; without it the command fails. Everything runs locally, so project details stay on your side.
What it helps with
- draft an ISTQB-aligned test strategy from a project description
- estimate QA effort and team capacity
- score an existing test plan across six quality dimensions
- spot flaky and ever-failing tests in a JUnit report
- assess QA process maturity against TMMi
Example prompts for the agent
- Build a test strategy for a React and Node.js web app
- Estimate the QA effort from this project description
- Review the quality of
test-plan.mdand list the weak spots
How to install qai-consultant-mcp
Platform CLI
claude mcp add qai-consultant-mcp -- uvx qai-consultant-mcpuvx runs the command named after the package; if the README names another one, use uvx --from qai-consultant-mcp <command>
Platform CLI
codex mcp add qai-consultant-mcp -- uvx qai-consultant-mcpuvx runs the command named after the package; if the README names another one, use uvx --from qai-consultant-mcp <command>
Config file
{
"mcpServers": {
"qai-consultant-mcp": {
"command": "uvx",
"args": [
"qai-consultant-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from qai-consultant-mcp <command>
Config file
{
"servers": {
"qai-consultant-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"qai-consultant-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from qai-consultant-mcp <command>
Platform CLI
gemini mcp add qai-consultant-mcp uvx qai-consultant-mcpuvx runs the command named after the package; if the README names another one, use uvx --from qai-consultant-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about qai-consultant-mcp
How do I install qai-consultant-mcp in Claude Code?
Run claude mcp add qai-consultant-mcp -- uvx qai-consultant-mcp. The server starts locally through uvx and needs no API keys.
What does qai-consultant-mcp do?
It connects your agent to a QA standards knowledge base and tools that generate a test strategy, risk register, effort estimation and test plan. It also reviews existing QA documents and analyses test results.
Does the server need API keys?
No. It runs locally through uvx and requires neither API keys nor Pinecone.
