Author verifiable eval records through a draft→review→revise→submit loop with enforced graders.
Connect Runs on your machine
claude mcp add evals-mcp-server -e EVALS_DATA_DIR=<EVALS_DATA_DIR> -e EVALS_REQUIRE_CONFIRMATION=<EVALS_REQUIRE_CONFIRMATION> -e EVALS_DEFAULT_LICENSE=<EVALS_DEFAULT_LICENSE> -e EVALS_CAPTURE_DIR=<EVALS_CAPTURE_DIR> -e MCP_LOG_LEVEL=<MCP_LOG_LEVEL> -- npx -y @cyanheads/[email protected]
{
"mcpServers": {
"evals-mcp-server": {
"command": "npx",
"args": [
"-y",
"@cyanheads/[email protected]"
],
"env": {
"EVALS_DATA_DIR": "<EVALS_DATA_DIR>",
"EVALS_REQUIRE_CONFIRMATION": "<EVALS_REQUIRE_CONFIRMATION>",
"EVALS_DEFAULT_LICENSE": "<EVALS_DEFAULT_LICENSE>",
"EVALS_CAPTURE_DIR": "<EVALS_CAPTURE_DIR>",
"MCP_LOG_LEVEL": "<MCP_LOG_LEVEL>"
}
}
}
}
{
"mcpServers": {
"evals-mcp-server": {
"command": "npx",
"args": [
"-y",
"@cyanheads/[email protected]"
],
"env": {
"EVALS_DATA_DIR": "<EVALS_DATA_DIR>",
"EVALS_REQUIRE_CONFIRMATION": "<EVALS_REQUIRE_CONFIRMATION>",
"EVALS_DEFAULT_LICENSE": "<EVALS_DEFAULT_LICENSE>",
"EVALS_CAPTURE_DIR": "<EVALS_CAPTURE_DIR>",
"MCP_LOG_LEVEL": "<MCP_LOG_LEVEL>"
}
}
}
}
{
"servers": {
"evals-mcp-server": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@cyanheads/[email protected]"
],
"env": {
"EVALS_DATA_DIR": "<EVALS_DATA_DIR>",
"EVALS_REQUIRE_CONFIRMATION": "<EVALS_REQUIRE_CONFIRMATION>",
"EVALS_DEFAULT_LICENSE": "<EVALS_DEFAULT_LICENSE>",
"EVALS_CAPTURE_DIR": "<EVALS_CAPTURE_DIR>",
"MCP_LOG_LEVEL": "<MCP_LOG_LEVEL>"
}
}
}
}
Replace the values in angle brackets with your own. Keys stay on your machine; we never see them.
Tools
This server runs locally, so its tools are listed by your MCP client once it starts.
Environment variables
| Name | Description |
|---|---|
EVALS_DATA_DIROptional | Root folder for record JSON. The store manages drafts/, submitted/, and exports/ subdirs under it. |
EVALS_REQUIRE_CONFIRMATIONOptional | When 'true', evals_submit_draft requests human confirmation before finalizing. |
EVALS_DEFAULT_LICENSEOptional | Default metadata.license applied when a draft omits one. |
EVALS_CAPTURE_DIROptional | Directory of framework-written tool-call captures; when set, captures EvalsIDs resolve to full dumps. |
MCP_LOG_LEVELOptional | Sets the minimum log level for output (e.g., 'debug', 'info', 'warn'). |
Details
| Package | @cyanheads/[email protected] (NPM, stdio) |
| Package | @cyanheads/[email protected] (NPM, streamable-http) |
| First published | 13 Oct 2026 |
| Last updated | 13 Oct 2026 |
| Registry name | io.github.cyanheads/evals-mcp-server |
More security servers See all
Listing data from the official MCP registry. Status and tools from NestDaddy's daily check. Browse all MCP servers.
