What this is
You need to look inside an Elasticsearch cluster, and standing up Kibana for five queries is not worth the trouble. Prismatic is one page of static files: you type the cluster URL, your browser talks to that cluster directly, and you get the console.
It works against both Elasticsearch and OpenSearch — the distribution is detected on connect, and everything it calls exists in both.
What it shows
- Overview. Cluster health, node count, documents, store size, JVM heap, shard states and pending tasks, refreshing on its own.
- Nodes. Heap, RAM, CPU, load and disk per node, with warning thresholds and the elected master marked.
- Indices. A sortable, filterable list; click through to mappings, settings and stats.
-
Shards. A grid of primaries and replicas per index. Click a
shard to run
_cluster/allocation/explainand read why it sits where it does, or why it is unassigned. - Search. A query builder with AND / OR / NOT groups — match, phrase, term, prefix, wildcard, exists, ranges, Lucene query strings — with field names completed from the live mapping, a preview of the DSL it generates, and a raw DSL mode when you want to write it yourself.
-
REST console. Any method against any endpoint, with the
diagnostic ones to hand (
_cat/*,_cluster/*,_nodes/hot_threads,_tasks, recovery, templates), per-tab history, and a confirmation step before DELETE.
Where your data goes
Nowhere except your cluster. There is no server component here, no proxy, no analytics and no third-party asset — the page is HTML, CSS and JavaScript served from this domain, and the content security policy in it locks scripts to the same origin.
A saved connection keeps only its name and URL, in
localStorage. Credentials are never written there: they
live in memory, and the optional "remember for this tab" box puts them
in sessionStorage, which the browser clears when you close
the tab. Console history is per tab and goes the same way.
What you need first
Because the requests come from your browser rather than from a server, the cluster has to allow cross-origin requests from this domain. The connection screen has a CORS setup help button with the settings for Elasticsearch and for OpenSearch. If you cannot change cluster settings, the same files can be served from the cluster's own origin behind your existing reverse proxy, and then no CORS is involved at all.
Basic auth and API keys are both supported. The source is MIT-licensed and lives at github.com/alexff91/prismatic.
Made by Aleksandr Fedorov at alftech.space. The cluster URL, the credentials and every request stay in this tab and go straight to your cluster: nothing is proxied through a server of mine. If a screen misreads your version of Elasticsearch or OpenSearch, or a number here disagrees with the cluster, write to a@alftech.space · buy me a coffee