Coval MCP Server
OfficialThe Coval MCP Server integrates Coval's AI evaluation platform into MCP clients, enabling you to manage and run evaluations of AI agents.
Evaluation Runs
List, get, and create evaluation runs — launch evaluations by combining an agent, persona, and test set, with options for tags, concurrency, and iteration count
Agents
List, get, create, and update AI agents (VOICE, OUTBOUND_VOICE, SMS, WEBSOCKET, CHAT) with configuration details like endpoints, phone numbers, and prompts
Test Sets & Test Cases
List, get, and create test sets (collections of test scenarios)
List, get, create, and update individual test cases, including single-turn or multi-turn conversations, expected behaviors, and metadata
Metrics
List available evaluation metrics and get detailed information about specific metrics
Personas
List and get simulated user personas used in evaluations, including voice, language, background sound, and behavior settings
Covi Delegation
Use
consult_covito ask read-only evaluation and analysis questions to Covi
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Coval MCP ServerShow my recent evaluation runs"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Coval MCP Server
The official Model Context Protocol server for Coval - the AI evaluation platform.
This MCP server allows AI assistants like Claude Desktop and Cursor to interact with Coval's evaluation APIs, enabling you to:
Launch and monitor evaluation runs
Manage AI agents and test sets
Retrieve evaluation metrics and results
Installation
npx @covalai/mcp-serverRelated MCP server: Trustwise MCP Server
Quick Start
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"coval": {
"command": "npx",
"args": ["-y", "@covalai/mcp-server"],
"env": {
"COVAL_API_KEY": "your_api_key_here"
}
}
}
}Cursor
Add to .cursor/mcp.json in your project:
{
"mcpServers": {
"coval": {
"command": "npx",
"args": ["-y", "@covalai/mcp-server"],
"env": {
"COVAL_API_KEY": "your_api_key_here"
}
}
}
}Remote Connection (Alternative)
{
"mcpServers": {
"coval": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://mcp.coval.dev/mcp",
"--header",
"X-API-KEY: ${COVAL_API_KEY}"
],
"env": {
"COVAL_API_KEY": "your_api_key_here"
}
}
}
}Get your API key from app.coval.dev/settings
Available Tools
Tool | Description |
| List all agents in your workspace |
| Get details of a specific agent |
| List evaluation runs |
| Get details of a specific run |
| Start a new evaluation run |
| List available test sets |
| Get test set details |
| List test cases in a test set |
| Add a test case to a test set |
| Get metrics for a run |
| List available personas |
| Delegate a read-only Coval evaluation question to Covi |
Example Usage
Once connected, you can ask Claude things like:
"Show me my recent evaluation runs"
"List all my agents"
"Run an evaluation of my customer-support-agent against the billing-inquiries test set"
"What are the metrics for run abc123?"
"Use
consult_covito analyze my latest failed run and recommend the most useful next test."
The same tools, including consult_covi, are available through both supported transports:
Remote Streamable HTTP:
https://mcp.coval.dev/mcpusing Clerk OAuth. This is the recommended connection for Codex, Claude, and other hosted MCP clients.Local stdio:
npx @covalai/mcp-serverwithCOVAL_API_KEY, for service accounts and local development.
Remote clients may continue to send X-API-Key during migration. OAuth access tokens terminate at
the MCP server and are never forwarded to Coval APIs or Sofia; the server exchanges verified Clerk
user and organization identity for Coval's existing managed per-user API key.
Remote OAuth operator requirements
Staging-enable prerequisite: the Clerk OAuth application MUST be configured to issue
JWT-format access tokens containing the selected organization in org_id or organization_id.
The server reads the organization only from that signature-verified token claim (Clerk's verified
OAuth auth object does not expose an organization id), so opaque oat_ tokens and
organization-less tokens are rejected with 401 and no fallback — every OAuth connection fails
until this is configured. This keeps organization selection bound to verified identity rather
than request parameters.
Enable Dynamic Client Registration for MCP clients that create their OAuth registration at connect
time, and keep the Clerk consent screen enabled so the user explicitly selects the organization
granted through user:org:read.
Do not publish the remote-connection release or repoint mcp.coval.dev until the backend identity
exchange and Sofia delegation endpoint are deployed, consult_covi succeeds through a real OAuth
connector, and the legacy API-key connector path has been regression-tested.
Development
# Install dependencies
npm install
# Build
npm run build
# Test locally with MCP Inspector
npm run inspector
# Run tests
npm testEnvironment Variables
Variable | Required | Default | Description |
| Stdio | - | Coval API key for the local stdio transport |
| No |
| API base URL |
| No | Derived from | Overrides the expected Sofia origin used to validate delegation URLs |
| Remote |
| Streamable HTTP listen port |
| Remote | - | Clerk publishable key used for OAuth metadata |
| Remote | - | Clerk server key used to verify OAuth access tokens |
| Remote OAuth | - | Internal credential used only for managed user-key exchange |
| No |
| Logging level |
Documentation
License
MIT
Support
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/coval-ai/mcp-server'
If you have feedback or need assistance with the MCP directory API, please join our Discord server