Dashboard

oblako dashboard launches a web UI built with the AWS Cloudscape Design System, the same components as the real AWS Console, at http://localhost:8000. It’s a read-and-act view over everything that’s running.

It listens on 127.0.0.1, so only your own machine can open it: it has no login, and its Notebook page runs Python against your services. --host makes it listen elsewhere, with a warning; do that only on a network you trust.

Pages

  • Services: status overview of all running services.

  • Notebook: a Python editor with syntax highlighting that runs code against all your local services.

  • Bedrock: chat playground powered by Ollama (or OpenRouter).

  • SageMaker: training jobs, endpoints, Docker images, and cleanup.

  • S3: bucket browser with object listing.

  • Athena: a SQL editor. Trino reads Iceberg and S3 Tables through the catalog, and DuckDB-Wasm runs in your browser against Parquet files in S3.

  • Glue: Data Catalog databases and tables (the same tables Athena sees), and PySpark jobs in the Glue 5 image.

  • Kinesis: streams, with a record writer and reader.

  • DynamoDB: table browser with an item viewer.

  • RDS / Aurora: instances and clusters, with database creation.

  • Step Functions: state machines, ASL JSON viewer, execution history, and a flow diagram.

  • CloudFormation: stacks deployed to the local CloudFormation, with their resources, outputs, and events.

  • Redshift: cluster list (management API), table list, and a SQL query editor with results.

  • Lambda: functions and layers, with function creation and invocation.

  • EC2: instances, each backed by a real container, with instance launch.

  • IAM: users, roles and policies, with assume-role and an access simulator.

  • AppConfig: applications, configuration profiles and deployments, with a feature-flag evaluator.

Notebook (JupyterLab)

oblako notebook launches JupyterLab with the kernel pre-wired to oblako, so you write the AWS code you normally would and it runs against your local services, no endpoint_url, no config:

oblako up                      # start the services
pip install 'oblako[notebook]'
oblako notebook                # JupyterLab on http://localhost:8888

import boto3
s3 = boto3.client("s3")        # transparently hits S3Proxy, no endpoint_url
s3.create_bucket(Bucket="from-notebook")

Under the hood the kernel gets AWS_ENDPOINT_URL_* for every service plus test credentials and S3 path-style/checksum config. The always-on Docker services work immediately; the in-process servers (CloudFormation, redshift-data, rds-data, bedrock-runtime) start on first use.