Services by Bitno
A lightweight, private and secure Services panel for VS Code, next to Terminal and Ports: Docker, Kubernetes, Postgres, Cassandra, Kafka, Redis, MongoDB, Elasticsearch and S3 in one tree, with the selected node's log, data or properties beside it, SSH terminals on your hosts, and editing of SQL, CQL, SQL++, Redis, MongoDB, Cypher, PartiQL and PromQL files.
Early development: this extension is still being built. Expect missing features, rough edges and breaking changes between versions.
Lightweight, private and secure by design
Built by a developer, for developers, from Serbia. Free to use, with no paid tier.
- Lightweight: Nothing connects until you expand it, and each client loads only the first time you use it.
- Private: No telemetry, no account, no cloud. It talks only to your Docker Engine, your clusters and the connections you save.
- Secure: Passwords stay in your OS keychain, never in settings, and the audit log is encrypted on your machine.
Services
- Docker and Docker Compose: a running Docker Engine, local or on a host reached through
ssh://; the docker compose CLI for compose actions.
- Postgres
- Cassandra and ScyllaDB
- Kafka
- Kubernetes: your kubeconfig's contexts.
- Redis, with Valkey, KeyDB and Dragonfly
- MongoDB
- Elasticsearch: the clusters Kibana shows logs from, logged in with a user and password
- S3 and the stores that speak its API, logged in with an access key or an AWS profile, SSO included
- SSH hosts: the
ssh client on the PATH.
Features
- Docker compose projects, containers, images, volumes and networks, refreshed from Docker's events, on this machine or on a host through
ssh:// as a Docker context reaches it
- Container logs that follow, with ANSI colors, timestamps, filter and wrap; stack frames link to the source
- Plain log lines colored by their level, the filter's matches marked, and the lines shown opened in an editor
- A real terminal into a running container, as its user or as root
- Live charts of a running container's CPU, memory against its limit, network and disk, and of all of them added up
- Charts of a container's or pod's Prometheus exporter, found by its annotations or port, with the metrics you pick
- Start, stop, restart, pause, remove and clean up, like
docker system prune
- Deploy, restart, stop and down compose projects, from the tree or from buttons in the compose file; Down keeps the project's volumes, and Down and Remove Volumes deletes them with it
- Compose Specification schema for completion and validation, with the YAML extension
- Postgres databases, schemas, tables, views, routines and sequences, with columns, keys and indexes
- Live charts of a Postgres database's connection pool, sessions, transactions, long transactions, deadlocks and cache hit ratio, warning when one needs a look
- Cassandra and ScyllaDB keyspaces, tables, materialized views, types and functions
- Live charts of a Cassandra node's requests and p99 latency, pending and blocked tasks, clients and key cache from
system_views
- Table data in pages, with sort and a
WHERE filter that completes column names, plus DDL and properties
- Copy a cell, a row or the rows shown, to paste into a spreadsheet, and export rows as CSV or JSON
- SQL and CQL consoles with highlighting and completion of tables and columns: Ctrl+Enter (Cmd+Enter on macOS) runs the statement under the cursor, and History puts back one that ran
- Kafka brokers, topics, partitions and consumer groups with their lag
- Live charts of a Kafka cluster's messages in by topic, consumer lag by group, and brokers and partitions at risk
- Kubernetes contexts, namespaces, pods, Deployments, StatefulSets and Services, updating as they change
- Pod logs per container, a shell in any container, CPU and memory charts from metrics-server, and port forwards to pods and Services
- Validation, completion and hover of Kubernetes manifests from your cluster's own schemas, CRDs included, for the current context once you expand it, with the YAML extension (Language Server for YAML by Bitno)
- Apply above each resource of a Kubernetes manifest, and Apply All for the file, to a cluster you expanded, after you confirm
- Follow a topic's messages like a log, and produce new ones
- Avro, Protobuf and JSON Schema records read as JSON through your Schema Registry, or Redpanda's built-in one
- Redis databases with keys grouped by prefix, each key's value by type and its TTL, and a command console
- MongoDB databases, collections, views and time series, with documents in pages filtered by a query document, indexes, statistics and a console of JSON commands
- Live charts of a MongoDB server's connections, operation latency, operations, documents examined against returned and WiredTiger cache, with its running operations and its replica set's members and lag
- Live charts of a Redis server's commands, clients against
maxclients, memory against maxmemory, fragmentation and hit ratio
- Elasticsearch data streams and indices, each followed as a log like Kibana's, with levels, errors in red and stack frames linked to the source
- Live charts of an Elasticsearch cluster's indexing and search rates and latency, heap, rejected requests and unassigned shards
- S3 buckets, folders and objects, with a preview of an object's start (gzip opened, JSON indented), its metadata, Download and Copy S3 URI
- Live charts of a MinIO deployment's drive space, drive latency, network and drive operations, with its servers and drives
- Databases and brokers running in Docker or Kubernetes show up without setup; others you add once and save
- TLS for Postgres, Cassandra, Kafka, Redis, MongoDB, Elasticsearch and S3 connections, verifying the certificate against the well-known CAs or a CA file, or accepting any certificate
- Connections of every service in one place, in folders you arrange as you like; drag connections between them
- Only the services you use show; search every saved connection and listed resource, and pin the ones you open often
- SSH hosts in folders, each a click from a terminal that uses your
ssh client and ~/.ssh/config, the hosts your ~/.ssh/config names offered beside them, and Connect to All opening a folder's hosts at once
- A connection that fails says what to fix, with Refresh and Edit Connection right on it
- An Audit tab on each connection, Docker and each Kubernetes context lists what you and AI assistants did there, the newest first, with what ran and how it went; filter it to yours or the AI's. It's kept encrypted on your machine, with passwords blanked
- Nothing connects until you expand a connection or a Kubernetes context; a filled dot at the right of each connection shows what's connected, an empty one what isn't
- Detach the tree and details into a window of their own
- AI assistants reach the connections you allow through a local MCP server, off until you turn it on: per connection, off, read or read and write, set with Set AI Access on its menu or in its form. On Postgres they list databases and tables, describe a table and run queries, read-only enforced by Postgres
Query files
- Languages:
.cql files open as CQL, .sqlpp and .n1ql as SQL++, .redis as Redis commands, .mongodb as MongoDB queries, .cypher, .cy and .cyp as Cypher, .partiql as PartiQL and .promql as PromQL, each with its comments, brackets and quotes. .sql files are VS Code's own SQL.
- Syntax errors: every one of them, SQL too, shows its syntax errors as you type, each underlined where it starts: a misspelt keyword, an unclosed bracket or an unterminated string.
- Colors: keywords, functions, types, names, strings, numbers and comments in their theme colors, from the same grammar that finds the errors. VS Code colors SQL itself.
- Formatting: Format Document lays SQL, SQL++, CQL and PartiQL out a clause per line, with nested queries indented, and Cypher a clause per line. MongoDB scripts are laid out with a pipeline's stages one per line, Redis files get one command per line, and PromQL queries are laid out the way Prometheus formats them. Your comments, indentation and line endings are kept.
- Outline and folding: statements show in the outline, the breadcrumbs and Go to Symbol in Editor, each named by what it does (
SELECT … FROM orders, CREATE TABLE users, db.orders.find), with a table's columns and a query's CTEs inside it. Multi-line statements, anything in brackets, block comments, #region comments and Redis's MULTI … EXEC fold.
- Completion of keywords and functions: the keywords that can come where you type, as
WHERE and ORDER BY after a table but not after SELECT, and each language's built-in functions where a function can be called, with how to call them and what they return. MongoDB files complete the shell's methods after db., a collection and a cursor; Redis files complete its commands.
- A file's database: Use Connection… binds a SQL or CQL file to a saved Postgres database or Cassandra keyspace, which the file's language status shows. The file then completes its tables after
FROM, JOIN, INTO and UPDATE, and the columns of the tables the statement names, or of a table or alias before a dot. Run above each statement runs it on that database and shows its rows in the details, where it can be cancelled; the file's first line names the database, and the first Run in a file without one asks for it.
- Status beside the language mode: a query file's
{} language status shows the language server's state ("Ready") or its work, and the file's database; it comes into the status bar when the server failed to start or work failed, and opens the Services output.
Coming soon
Requirements
- VS Code 1.145 or later.
- For Docker: a running Docker Engine, and the
docker compose CLI for compose actions.
- For SSH hosts, and Docker on a host through
ssh://: the ssh client on the PATH.
- For Kubernetes manifests' validation, completion and hover: Language Server for YAML by Bitno. Everything else works without it, and Services offers to install it when you open a manifest of a connected cluster.
Getting started
Open the Services panel next to Terminal, or drag it wherever you like. Docker appears as soon as the engine is reachable. To connect to a database or broker that isn't running in Docker, click + in the panel's title and pick the service: servers running in Docker are offered filled in, or fill in a new connection. Right-click a service for New Folder to organize what you save.
Settings
| Setting |
Default |
Description |
bitno.services.docker.host |
empty |
Docker Engine address (unix://, npipe://, tcp://). Empty uses the Docker context (docker context use, as Colima sets) or DOCKER_HOST, then the default socket. |
bitno.services.logs.tail |
1000 |
Recent log lines loaded before following new ones. |
bitno.services.connections |
[] |
Saved connections, written by Add Connection. |
bitno.services.connectionFolders |
[] |
Folders created with New Folder, kept while they're empty. |
bitno.services.groupBy |
folder |
folder keeps every service's saved connections under Connections, in your folders; service shows each service with its own. |
bitno.services.visibility |
{} |
Services shown (true) or hidden (false) by id; unlisted ones show unless they save connections and have none. |
bitno.services.ai.enabled |
false |
Runs the MCP server AI assistants reach the services through, on 127.0.0.1 with a token. VS Code finds it; Copy MCP Config gives it to other clients. |
bitno.services.docker.aiAccess |
off |
What AI may do on Docker: off, read or write. Set AI Access on Docker sets it. |
bitno.services.kubernetes.aiAccess |
{} |
What AI may do on each kubeconfig context, by name; unlisted ones are off. Set AI Access on a context sets it. |
- Your Docker Engine, your kubeconfig's clusters, and the connections you save, a connection's Schema Registry included. SSH hosts keep no password; your
ssh client asks for it.
- Query files connect to nothing while you edit them; a file's tables load only from the database you bind it to.
- AI assistants reach nothing unless you turn on the MCP server and allow a connection. What they read then goes to your AI client, which may send it to its model's provider.
- Nothing else.
Feedback
Found a bug, or missing something? Open an issue. Questions and ideas go to Discussions. Blunt feedback is welcome.
Docker, Kubernetes, PostgreSQL, Cassandra, ScyllaDB, Redis, MongoDB, Kafka, Amazon S3, MinIO, Couchbase, Neo4j and Prometheus are trademarks of their respective owners.
| |