dockndevai — MCP server suite
Let agents run real infrastructure, read-only by default
Start new thread

dockndevai — MCP server suite - Let agents run real infrastructure, read-only by default

by•
Eleven open-source, safe-by-default Model Context Protocol (MCP) servers that let AI agents operate real infrastructure — Kubernetes, Kafka, ClickHouse, Debezium, OCI, Azure, Azure DevOps, Keycloak, Percona PostgreSQL, Grafana — plus RAG over your own documents. Read-only by default, allowlisted, audited.

Add a comment

Replies

Best
Hunter
📌
Hey Product Hunt! I'm Ankit. I kept hitting the same wall: agents are great at ops work, but wiring one up to Kubernetes or a production database means handing it the keys — a non-starter for anything real. So I built these the opposite way. Every server starts read-only; the write and delete tools aren't even registered until you raise the access mode on purpose, and destructive actions need a second explicit opt-in. Same governance model across all 11: resource allowlists, protected resources, dry-run, JSON audit logging, and secret redaction. Covered so far: Kubernetes, Kafka, ClickHouse, Debezium, OCI, Azure, Azure DevOps, Keycloak, Percona PostgreSQL, Grafana — plus one for RAG over your own docs + durable agent memory. All open-source, on npm + the official MCP registry. Would genuinely love your feedback — especially on the access-mode design, and which system you'd want covered next.