Evalgate runs on your own machine — the client starts it, so there is no endpoint to ping. 22 installs a week from pypi.
Statistical checks an agent runs before trusting an AI eval number (is #1 real, judge bias, more).
The linked repository no longer exists on GitHub — it was deleted or made private.
This one has been quiet for a while
Quiet is not dead — but it is worth knowing when it wakes up, or when someone else takes it over. We watch the repository and tell you either way.
Connect this server
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add evalgate -- uvx eval-integrity{
"mcpServers": {
"evalgate": {
"args": [
"eval-integrity"
],
"command": "uvx"
}
}
}[mcp_servers.evalgate]
command = "uvx"
args = ["eval-integrity"]{
"mcpServers": {
"evalgate": {
"args": [
"eval-integrity"
],
"command": "uvx"
}
}
}{
"mcpServers": {
"evalgate": {
"args": [
"eval-integrity"
],
"command": "uvx"
}
}
}Alternatives to Evalgate
same job, measured the same wayCheck an agent's INAM reputation before trusting it, and sign a receipt when work is done.
MCP proxy that checks every tools/call against Agent Trust policy, with evidence, before it runs.
Verify x402 payment endpoints before an AI agent pays: scam scan, on-chain checks, trust scores.
Reputation scoring for AI agent wallets on Base. Trust scores, fraud checks, x402.
Safety checks an agent runs before it acts: package malware, destructive command, secret leak.
Statistical regression testing for LLM agents: p-value, effect size, and CI on behavior change.
12 tools that let AI agents check trust scores and reviews before recommending any software tool.
Check any MCP server or agent skill before installing: upkeep, real adoption, and OSV advisories.
Evalgate — questions
Answers built from our own checks of this server.