# EvalGate

> EvalGate is a freemium MCP server for ai models and llm tooling. Run AI evaluations, record agent traces and gate releases from your coding agent. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Freemium
- **Starting price:** $99 a month
- **Platforms:** Web, API
- **Website:** https://www.evalgate.com/docs/platform/mcp-integration
- **MCP Registry name:** com.evalgate/evalgate
- **Categories:** AI models and LLM tooling, developer tools
- **Listing updated:** 2026-11-15

## About

EvalGate's server exposes its evaluation platform to agents through a permission-filtered endpoint at /api/mcp. Anonymous clients can discover the protocol and read-only schemas, but cannot run product tools; everything else needs an API key as a bearer token. Each write tool lists a minimum role and a required scope, such as eval:write or runs:write, and says how to verify the effect, for example reading back an ID or polling runs_list and run_explain.

Authoring tools include evaluations_create, test_cases_add (locked evaluations refuse), runs_start (after readiness and spend guards) and traces_create, with a reminder to redact content and check retention. The docs also describe repository improvement campaigns in which an agent can compile an optimisation contract and open a draft PR, but cannot approve a hosted contract, adopt a candidate or merge. Merging is never automated.

A separate docs endpoint is public. The older /api/mcp/tools and /api/mcp/call belong to a different proprietary HTTP Tool API. Plans start free with 1,000 test results a month, and extra results cost $1 per 1,000.

## Features

- **Evaluation authoring:** Create evaluations and add reviewed test cases from an agent.
- **Guarded runs:** Start runs only after readiness and spend checks, then poll for results.
- **Trace recording:** Record agent traces with a reminder to redact content.
- **Scoped access:** Each tool has a minimum role and a required scope.
- **Human approval limits:** Agents cannot approve hosted contracts, adopt candidates or merge PRs.
- **Public docs endpoint:** A separate docs server needs no key.

## Good for

- Create an evaluation from a coding assistant
- Start an evaluation run and read why it failed
- Record an agent trace for later review
- Gate a release on test evidence

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http evalgate https://www.evalgate.com/api/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://www.evalgate.com/api/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "evalgate": {
      "url": "https://www.evalgate.com/api/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "evalgate": {
      "type": "http",
      "url": "https://www.evalgate.com/api/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "evalgate": {
      "serverUrl": "https://www.evalgate.com/api/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "evalgate": {
      "type": "http",
      "url": "https://www.evalgate.com/api/mcp"
    }
  }
}
```

## Tools

- `evaluations_create`: Creates an evaluation and returns its ID.
- `test_cases_add`: Adds a reviewed test case; locked evaluations refuse.
- `runs_start`: Starts a run after readiness and spend guards.
- `traces_create`: Records an agent trace.
- `runs_list`: Lists runs so progress can be polled.
- `run_explain`: Explains the outcome of a run.

## FAQ

### Can anonymous clients run tools?

No. Anonymous clients can discover the protocol and schemas but cannot execute product tools.

### How do I authenticate?

Send an EvalGate API key as an Authorization bearer header.

### What does it cost?

A free Developer plan includes 1,000 results a month; Team is $99 for 100,000 and extra results are $1 per 1,000.

### Can an agent merge a pull request?

No. The docs say merging is never automated by EvalGate.

### Whose model costs apply?

Your own provider bill, with no EvalGate markup.

### Is EvalGate free?

EvalGate has a free plan, with paid plans for more usage or features. Paid plans start at $99 a month.

### How do I add EvalGate to Claude Code?

Run this command in your terminal: claude mcp add --transport http evalgate https://www.evalgate.com/api/mcp

### What are the best alternatives to EvalGate?

The closest alternatives to EvalGate are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Concord MCP](https://ai-powered.com/mcp/concord-mcp): Concord MCP is a freemium MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/evalgate
