Hotdata CLI
Command line interface for Hotdata.

Query, search, and join your data from one place — external databases, APIs, cloud storage, Iceberg catalogs, and files you upload — with plain SQL and a few commands.

Install

brew install hotdata-dev/tap/cli        # Homebrew
cargo install --path .                  # from source (requires Rust)

Or grab a binary from Releases. Stay current with hotdata manage upgrade; enable tab completion with hotdata manage completions bash|zsh|fish.

Quickstart

hotdata auth login                            # or: hotdata auth register
hotdata databases create --catalog demo
hotdata databases load --catalog demo --table trips \
  --url https://d37ci6vzurychx.cloudfront.net/trip-data/yellow_tripdata_2024-01.parquet
hotdata query "SELECT count(*) FROM demo.public.trips"

The core loop: create an instant database, put data in it, query it with PostgreSQL-dialect SQL. Everything else builds on that.

Getting your data in

Upload a file directly — csv, json, or parquet:

hotdata databases load --catalog demo --table listings --file ./listings.csv

The format comes from the extension; pass --format csv|json|parquet when the extension is missing or misleading. json is read whatever shape it arrives in — an array of objects, a pretty-printed document, or one object per line.

A load replaces the table by default. --mode append adds rows to an existing table instead:

hotdata databases load --catalog demo --table listings --file ./more-listings.parquet --mode append

Import from an external source — Postgres/MySQL, S3/GCS buckets, Iceberg, Kafka, ~150 API services — via a datasource (ds_…) and a saved ingest (ing_…):

hotdata ingest sources types                  # browse source types and families
hotdata ingest sources fields sql             # config/credentials/selector a family takes
hotdata ingest sources add                     # create a datasource (prompts, or --config @src.json)

hotdata ingest create --source "prod postgres" --table orders --database-id db_123
hotdata ingest logs <ing_…>                    # attempts for an ingest, newest first
hotdata ingest run <run_…> --wait              # show one attempt; exits 0 done / 1 failed / 2 in flight

ingest sources update-config rotates credentials; ingest pause|resume|schedule control a scheduled ingest. Data lands in an instant database — query it like any other.

Query and explore

hotdata databases tables list                  # every queryable table
hotdata databases tables show <table>          # columns and types
hotdata query "<sql>" [-o table|json|csv]      # HotSQL (PostgreSQL dialect)

Write SQL in another dialect and the server transpiles it to HotSQL — --dialect accepts hotsql (default), duckdb, postgres, snowflake (read-only for a non-default dialect):

hotdata query "SELECT IFF(n > 0, 'pos', 'neg') FROM t" --dialect snowflake

Long queries go async and print a query_run_id — poll with hotdata query status <id> (exit 0 done / 1 failed / 2 running / 3 done but the printed result is a truncated preview). Re-fetch past results with hotdata databases results get <result-id>; browse history with hotdata databases queries list.

Join across databases

Attach another instant database and join its live tables directly, no copying:

hotdata databases attach prod-replica --alias prod
hotdata query "SELECT t.id, o.total FROM demo.public.tickets t
               JOIN prod.public.orders o ON o.ticket_id = t.id"

The attached database is read-only here: loads still go to your own database, and detach withdraws visibility without deleting anything. --alias is required when the other database kept the stock default catalog name.

Search

Create an index once, then search server-side. Vector search auto-embeds the column and the query — no embedding keys or client setup:

hotdata search create trips_notes --type text --from demo.public.trips --column notes
hotdata search "airport surcharge dispute" --index trips_notes

Use --type vector for semantic search. Indexes resolve in the active database (hotdata databases use <id>); pass -d/--database <id> to target another. Bring your own model with hotdata search embeddings add.

Use it from scripts and agents

• Every listing command takes -o json|yaml; query status and ingest run expose script-friendly exit codes.
• Authenticate non-interactively with --api-key, or HOTDATA_API_KEY in the environment or a .env file.
• hotdata manage skills install installs bundled agent skills — Markdown playbooks that teach AI coding agents (Claude Code and friends) the full CLI.
• hotdata databases context push|show DATAMODEL stores your data model as shared, server-side Markdown so humans and agents query with the same map.

Getting help

File a support ticket without leaving the terminal:

hotdata support report -m "Queries against work_abc have been timing out for an hour" --subject "Queries timing out"
hotdata support report --logs ./stderr.txt --context env=staging

Omit -m in an interactive terminal to compose the report in $EDITOR instead. Replies go to the email on your HotData account.

Commands

The full command surface. The top level has nine groups — auth, workspaces, databases, query, jobs, ingest, search, manage, and support. Run hotdata <command> --help for full flags on any command.

Configuration

Config lives at ~/.hotdata/config.yml (profile-keyed). Environment variables:

API-key precedence, lowest to highest: config file → HOTDATA_API_KEY → --api-key.

Development

cargo build && cargo test

Release process: see docs/RELEASING.md.