About DevOpsPad
DevOpsPad is a desktop workspace for the infrastructure a small team runs every day: a few databases, a cache, a Kafka cluster and one or more K3s or Kubernetes clusters.
It started as a Navicat-style database client and grew into a local-first control panel for small deployments. The rule that shaped it: reading is safe by default, and every write is a plan you review first.
Who it is for
- Developers who need to look at production data without risking a stray
UPDATE. - Small teams running K3s on a handful of machines, without Argo CD, Rancher or a platform team.
- On-call engineers who want resource details, events and Pod logs next to the database they are debugging.
What it deliberately does not do
- It does not edit data, change schemas or run migrations. The workbench is read-only.
- It does not run a server, agent or in-cluster controller.
- It does not let the AI assistant execute anything. Suggestions are inserted into the editor; you run them.
- It does not upload query results, logs or credentials anywhere.
Features that are not built yet, such as editable grids, Kafka message browsing, Redis Cluster and live log streaming, are listed on each connector page.
How it is built
DevOpsPad is a Wails app: a Go backend with a React, TypeScript and CodeMirror interface. Connectors use the gotomicro/ego component ecosystem where it fits, and upstream drivers where it does not. Read more on the engineering blog.
Get in touch
Source code and releases live on GitHub.
Frequently asked questions
What is DevOpsPad?
DevOpsPad is a desktop app for browsing MySQL, Redis, MongoDB, Kafka, ClickHouse, Kubernetes and K3s read-only, plus an Infra console that releases, scales, restarts and rolls back K3s Deployments after a reviewed dry-run plan.
Can DevOpsPad modify my databases?
It is designed not to. MySQL queries run in server-side read-only transactions, ClickHouse with readonly=1, Redis through a read-command allow-list and MongoDB through find only. MySQL DDL is not reliably blocked by a read-only transaction, so connect with a read-only account for a hard guarantee.
Where are my passwords stored?
In the operating system credential store (the macOS keychain). Connection metadata is a local JSON file. Saved secrets are never returned to the interface, and saving fails rather than falling back to plaintext.
Does DevOpsPad send my data anywhere?
Only to the servers you connect to, to the AI provider you configure, and to GitHub to check for updates. There is no DevOpsPad backend, account or analytics.
Which AI models does it work with?
Any provider that implements the OpenAI-compatible /chat/completions endpoint, including local servers on your own machine. You supply the base URL, model ID and an optional API key.
Do I need to install anything on my servers or clusters?
No. DevOpsPad connects directly or through an SSH tunnel. There is no agent on hosts and no controller in the cluster.