MCP Selection Lab
Benchmark how well an agent picks the right MCP tool from metadata alone
- Platforms
- Web, API
- Runs
- Remotely
- Version
- 0.5.2
- Listing updated
- today
Similar MCP servers
About MCP Selection Lab
MCP Selection Lab is an MCP server for developer tools. It runs as a hosted remote server, so there is nothing to install locally.
The Agent Trust Lab site behind this server is a workspace for exploring MCP providers, inspecting versioned evidence per capability, and reviewing task-specific options before use. The server's own focus is tool selection: routing, collision, abstention and holdout checks that use only metadata.
The site draws careful lines. Its free scan generates test cases from tool descriptions, so it does not measure natural-language or real-model routing accuracy, and it does not call a provider's business tools. Data-quality grades come from five Registry criteria, are calibrated within its own dataset, and are explicitly not safety, accuracy, availability or reliability ratings. Evidence grades do not certify safety or rank providers.
The catalogue is indexed from the Official MCP Registry, and endpoints have not been checked.
Install
Run in a terminal
claude mcp add --transport http mcp-selection-lab https://agenttrustlab.com/mcpAdd to Settings → Connectors → Add custom connector
https://agenttrustlab.com/mcpAdd to ~/.cursor/mcp.json
{
"mcpServers": {
"mcp-selection-lab": {
"url": "https://agenttrustlab.com/mcp"
}
}
}Add to .vscode/mcp.json
{
"servers": {
"mcp-selection-lab": {
"type": "http",
"url": "https://agenttrustlab.com/mcp"
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"mcp-selection-lab": {
"serverUrl": "https://agenttrustlab.com/mcp"
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"mcp-selection-lab": {
"type": "http",
"url": "https://agenttrustlab.com/mcp"
}
}
}Using several servers? Build one config file with the MCP config generator.
What it does
- Metadata-only routing check
- Tests whether tool descriptions let an agent choose correctly.
- Collision checks
- Looks for tools whose descriptions overlap.
- Abstention and holdout
- Includes checks for when an agent should decline and for held-out cases.
- Registry data grades
- Grades listing data quality from five registry criteria.
- Versioned evidence
- Shows evidence by capability and version.
Good for
- Check whether your MCP tool descriptions are distinguishable
- Compare MCP providers before adopting one
- Review registry changes between snapshots
Pricing
Pricing unknown
A free scan is offered; other prices not published
Prices change. Check agenttrustlab.com before you buy.
MCP Selection Lab FAQ
Does it call my tools?
No. The scan inspects public discovery metadata and doesn't call business tools.
Does it measure real model accuracy?
No. The free scan generates cases from tool descriptions.
Are the grades safety ratings?
No. They describe registry data quality only.
How do I add MCP Selection Lab to Claude Code?
Run this command in your terminal: claude mcp add --transport http mcp-selection-lab https://agenttrustlab.com/mcp
What are the best alternatives to MCP Selection Lab?
The closest alternatives to MCP Selection Lab are REA, Context7, Chrome DevTools MCP, Codebase Memory and GitHub.
Where this comes from
- Description: written from the project's README and the project's website.
- Install steps: from the official MCP Registry entry.


