From host inventory to Kubernetes in-pod debug, from PostgreSQL replication to 200 GB+ HA — muppy.io operates the full stack from a Python + PostgreSQL backoffice, AI agent-driven via MCP. Zero hyperscaler dependency.
Concretely reducing human action surface
Sovereign (OVH, Scaleway, Infomaniak) + hyperscalers
Multi-DC switchover without healthcheck reconfiguration
All in production, all verified in code
Two specific offers
One tool to provision, deploy, debug and monitor your stacks — CI/CD, K8s Tag Mapper, in-pod debug, full Helm journal.
Regain control of your on-premise infrastructure. Zero hyperscaler dependency, full audit trail, extensible by your teams.
Capabilities that reduce human action surface during incidents, migrations and upgrades
Iterative detection + kill of blocking locks, schedulable via cron
Your databases stay available even under load: blocking locks are detected and removed automatically without human intervention.
Python GUI, topology-aware, Pushover push with no third-party service
Smartphone alert in seconds via Pushover; distinguishes primary vs. secondary failure without reconfiguration at switchover.
Rsync + 7 Prometheus metrics + transparent switchover
DC switchover is transparent at 200 GB+ filestore: rsync resumes in the correct direction without touching the configuration.
Cloudflare, GitLab, Azure DevOps — always up-to-date from their source
Your firewall rules stay consistent with IPs published by major SaaS providers — automatically updated from their official source.
TTL per UFW rule — automatic expiry, hourly cleanup cron
Temporary access expires on schedule — no more orphaned rules accumulating and eroding security.
Stop → backup → migrate → restart, traced per database
Each upgrade follows a traceable protocol per database — clean stop, backup, migration, restart — triggerable from CI/CD without SSH.
Semver git tag → selected K8s profile + automatic Helm deploy
A simple semver git tag triggers the right K8s profile selection and Helm deployment — zero manual clicks between push and production.
Filter by day, live link to IMQ log from each task_run
Each scheduled task is linked to its IMQ execution log in one click — Python logger + remote stdout/stderr visible from the UI without SSH.
K8s + Traefik + cert-manager + PG in 2 chained wizards
A complete Kubernetes cluster is operational in two chained wizards — onboarding a new server to production takes minutes.
VS Code in the browser in 1 minute on any host
A complete VS Code environment installs in 1 minute on any existing host — debug from the browser with no dedicated infrastructure.
Toggle → VS Code injected into the production pod (same env)
A toggle reconfigures the pod with code-server and the same mounts and variables — the bug is reproduced in its exact environment, not a copy.
Private + public route per pod, Zero Trust WebAuthn passkey 1-toggle
Each pod exposes a private route and a WebAuthn passkey-protected public route in one toggle — no manual Ingress config, no risk of oversight.
3-level diff + exact Helm command + stdout/stderr + exit code
Each Helm operation is logged with its 3 diff levels and stdout/stderr — post-mortem of a K8s incident without digging through cluster logs.
Python + PostgreSQL backoffice (Odoo Community Framework) with MCP exposed — every resource is an ORM record. Each Fabric operation captures Python log + subprocess stdout/stderr. No SSH needed to debug what happened.
Muppy is a sovereign all-in-one platform that integrates Kubernetes, managed PostgreSQL, DNS, TLS, high availability and CI/CD in a unified interface. Unlike Portainer (container management) or Ansible (configuration automation), Muppy covers the entire ops lifecycle — from provisioning to production — and can be driven by an AI agent via MCP.
Yes. Muppy is designed to run on your own hardware with no dependency on hyperscalers. It supports OVH, Scaleway, Infomaniak, Ikoula, Contabo and Dell ECS, as well as international hyperscalers (AWS, Azure, Google Cloud, Cloudflare, DigitalOcean). Full audit trail, extensible by your teams.
Muppy includes several HA mechanisms for PostgreSQL: automatic removal of blocking locks (PG Lock Auto-removal), topology-aware Smart Healthchecks with Pushover alerts, 200 GB+ HA Filestore with rsync and transparent switchover, and TTL-based ephemeral firewall rules. Datacenter failover is transparent with no reconfiguration required.
Yes. Muppy exposes an MCP (Model Context Protocol) surface of approximately 35 methods covering provisioning, deployment, command execution, file reads, journalctl logs and systemd control. A Claude, Cursor or any compatible MCP client can manage your servers directly, without opening a terminal.
The K8s Tag Mapper turns a semver git tag into a full Kubernetes deployment: it automatically selects the right K8s profile and triggers the corresponding Helm deployment. Zero manual clicks between the push and production. Every operation is logged with a 3-level diff, stdout/stderr and exit code for post-mortem analysis without digging through cluster logs.
Muppy Enterprise installs on-premise on your hardware. €490/month for the control plane. No proprietary ERP licence.