Safe Kaggle MCP runs on your own machine — the client starts it, so there is no endpoint to ping. 61 installs a week from pypi. Last commit 16 Jun 2026.
Safety-first Kaggle MCP — confirm-gated destructive ops, submission budget, injection-fenced.
Today is the operative word: we check Safe Kaggle MCP every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
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 kaggle-mcp -- uvx safe-kaggle-mcp
{
"mcpServers": {
"kaggle-mcp": {
"args": [
"safe-kaggle-mcp"
],
"command": "uvx"
}
}
}
[mcp_servers.kaggle-mcp]
command = "uvx"
args = ["safe-kaggle-mcp"]
{
"mcpServers": {
"kaggle-mcp": {
"args": [
"safe-kaggle-mcp"
],
"command": "uvx"
}
}
}
{
"mcpServers": {
"kaggle-mcp": {
"args": [
"safe-kaggle-mcp"
],
"command": "uvx"
}
}
}
This one needs environment variables set before it will start:
KAGGLE_USERNAME (Kaggle account username (legacy scheme; pair with KAGGLE_KEY). Optional — read-only tools work without credentials.), KAGGLE_KEY (Kaggle API key, the 40-char 'key' from kaggle.json (legacy scheme; pair with KAGGLE_USERNAME).), KAGGLE_API_TOKEN (Modern Kaggle KGAT_... token (alternative to KAGGLE_USERNAME/KAGGLE_KEY). Optional.), KAGGLE_MCP_ENABLE_DESTRUCTIVE (Set to 1 to expose delete tools. Default 0 (off).), KAGGLE_MCP_ENABLE_PUBLISH (Set to 1 to allow making datasets/kernels public. Default 0 (off).), KAGGLE_MCP_SUBMISSION_CAP (Daily per-competition submission budget. Default 5.).
The author declared them in the registry entry; get the values from the project itself.
Safety-first Terraform: confirmation-gated apply/destroy, risk & cost analysis, drift detection
lu.ma events MCP server with draft+confirm safety, audit logging, and prompt-injection scrubbing.…
AI security firewall & DLP proxy protecting API keys and blocking destructive commands.
Safe Google Workspace admin for AI agents: read-only by default, confirm-gated deletes, audit log.
CSV-first analytics for Google Ads + Meta Ads: reporting, anomaly detection, budget optimization.
Autodesk Construction Cloud via APS — projects, issues, RFIs, documents, submittals.
AI agent security: 7 MCP tools for injection detection, PII scanning, command safety, DLP.
Email safety MCP server. Detects phishing, prompt injection, CEO fraud for AI agents.
Answers built from our own checks of this server.