# geminiupgradeqa-mcp

Run evaluations for Gemini model upgrades, compare prompt responses, and spot regressions before a release. The remote MCP service also exports evaluation records for audits.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI
- Author: clauxel
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/clauxel/gemini-upgrade-qa-mcp
- Updated: 2026-05-20
- HTML: https://affpapa.org/en/nevibecode/mcp/geminiupgradeqa-mcp

## What it does

geminiupgradeqa-mcp is a paid remote MCP service for evaluating Gemini upgrades. It can run upgrade checks, compare model responses to prompts, flag regressions, and enforce blocking rules. Results can be issued as an evaluation receipt or exported for audit use.

Its tools include `run_gemini_upgrade_eval`, `compare_prompt_outputs`, `detect_model_regression`, `issue_upgrade_receipt`, and `export_eval_audit`. The checks focus on Gemini model upgrades and prompt behavior, rather than general project testing. Access is paid.

For Claude Code, add the remote server with `claude mcp add --transport http geminiupgradeqa-mcp https://geminiupgradeqa.clauxel.com/mcp`. The listed integrations also include Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, and Gemini CLI.

## Requirements
- Remote server: `https://geminiupgradeqa.clauxel.com/mcp`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http geminiupgradeqa-mcp https://geminiupgradeqa.clauxel.com/mcp
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add geminiupgradeqa-mcp --url https://geminiupgradeqa.clauxel.com/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "geminiupgradeqa-mcp": {
            "url": "https://geminiupgradeqa.clauxel.com/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "geminiupgradeqa-mcp": {
            "type": "http",
            "url": "https://geminiupgradeqa.clauxel.com/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http geminiupgradeqa-mcp https://geminiupgradeqa.clauxel.com/mcp
```

## When to use it
- evaluate prompts across a Gemini upgrade
- compare model responses between versions
- flag regressions in Gemini output
- block a release when rules trigger
- export evaluation records for an audit

## Example prompts
- Run an evaluation of these prompts before the Gemini upgrade.
- Compare Gemini responses across these model versions and flag regressions.
- Create an evaluation receipt and export the audit record for this upgrade.

## Similar
- [google-agents-cli-eval](https://affpapa.org/en/nevibecode/skills/google-agents-cli-eval.md) — skill · 511400 installs
- [verify-and-stop](https://affpapa.org/en/nevibecode/skills/verify-and-stop.md) — skill · 104900 installs
- [firecrawl-qa](https://affpapa.org/en/nevibecode/skills/firecrawl-qa.md) — skill · 31400 installs
- [argent-screenshot-diff](https://affpapa.org/en/nevibecode/skills/argent-screenshot-diff.md) — skill · 25000 installs
- [mo-qa](https://affpapa.org/en/nevibecode/skills/mo-qa.md) — skill · 23200 installs
- [dogfood](https://affpapa.org/en/nevibecode/skills/dogfood.md) — skill · 11300 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md