MCP server that lets AI agents understand any local dataset (CSV, Parquet, JSON, Excel) without reading raw rows — types, nulls, ranges and data-quality flags at ~100x less context.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add data-profiler -- uvx mcp-data-profiler~/.codex/config.toml
[mcp_servers.data-profiler]
command = "uvx"
args = ["mcp-data-profiler"]
# codex mcp add data-profiler -- uvx mcp-data-profiler~/.cursor/mcp.json
{
"mcpServers": {
"data-profiler": {
"command": "uvx",
"args": [
"mcp-data-profiler"
]
}
}
}.vscode/mcp.json
{
"servers": {
"data-profiler": {
"type": "stdio",
"command": "uvx",
"args": [
"mcp-data-profiler"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"data-profiler": {
"command": "uvx",
"args": [
"mcp-data-profiler"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"data-profiler": {
"command": "uvx",
"args": [
"mcp-data-profiler"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"data-profiler": {
"command": "uvx",
"args": [
"mcp-data-profiler"
]
}
}
}