lakefront 0.9.3

Lakehouse Observability Platform

Lakefront

A terminal-based lakehouse observability platform for exploring and managing data sources from your command line.

---

Installation

pip install lakefront

Or with uv:

uv add lakefront
# or install as a standalone tool
uv tool install lakefront

---

Quickstart

The fastest way to see Lakefront in action is the built-in demo:

lakefront init
lakefront demo

This seeds a sample project with two CSV datasets and opens it straight in the TUI.

To set up your own project instead:

# 1. Initialise ~/.lakefront
lakefront init

# 2. Create a project
lakefront projects create my-project -d "My first project"

# 3. Attach a data source (local Parquet or CSV)
lakefront projects source add -p my-project -n orders --uri /data/orders.parquet

# 4. Open the TUI
lakefront ui --project my-project

Running lakefront ui without --project opens a navigation screen listing all your projects.

---

About

Working with lakehouse data — Parquet files on local disk or S3, DuckDB queries, materialized views — usually means jumping between tools, writing throwaway scripts, or wrestling with heavyweight UIs. Lakefront puts it all in one place: a fast TUI and CLI for data engineers who live in the terminal.

Problems it solves:

• No single tool for lakehouse exploration — Lakefront combines DuckDB-powered SQL querying, S3 source management, and dataset browsing in one cohesive interface.
• Configuration sprawl — profiles let you switch between environments (local dev, staging, production S3) with a single command, keeping credentials out of your scripts.
• Context switching — instead of firing up a Jupyter notebook or a GUI just to peek at a Parquet file, you stay in the terminal.

---

Examples

Initialise Lakefront

Bootstrap the ~/.lakefront directory structure and create a default profile:

lakefront init

Config Management

# List all profiles
lakefront config list

# Show config directories and paths
lakefront config info

# Create a new profile
lakefront config create --profile staging

# Inspect a profile's current settings
lakefront config inspect --profile staging

# See which profile is active
lakefront config get-active

# Switch to a different profile
lakefront config set-active --profile staging

Secrets (S3 access keys etc.) can be written to the TOML profile or set via environment variables instead:

export LAKEFRONT_S3__ACCESS_KEY=...
export LAKEFRONT_S3__SECRET_KEY=...

Project Management

Projects are the top-level organisational unit in Lakefront. Each project lives in its own directory under ~/.lakefront/projects/ and can be pinned to a config profile.

~/.lakefront/projects/
└── my-project/
    ├── project.toml      ← metadata + pinned profile
    └── results/          ← analysis outputs

# List all projects
lakefront projects list

# Create a new project
lakefront projects create my-project -d "EDA on S3 parquet" -p staging

# Inspect a project
lakefront projects inspect my-project

# Delete a project (prompts for confirmation)
lakefront projects delete my-project
lakefront projects delete my-project --yes

Source Management

Data sources are attached to a project and point to a local path or S3 prefix. The source type is inferred from the URI scheme — bare paths and file:// URIs resolve to local files, s3:// to S3, and other known schemes are passed through as-is.

# Add a source (scheme inferred from the URI)
lakefront projects source add -p my-project -n raw --uri s3://bucket/raw/
lakefront projects source add -p my-project -n local --uri /data/parquet/

# Remove a source
lakefront projects source remove -p my-project -n raw

---

TUI (Terminal User Interface)

The interactive TUI provides a rich, multi-pane interface for exploring and analyzing data without leaving the terminal.

Project Screen

The main project workspace with a three-pane layout:

Left Pane — Data Sources:

• Browse all attached sources with expandable tree view
• View column names and data types inline
• Quickly navigate between sources
• P: Load statistical profile for the selected source in the right pane

Center Pane — SQL Editor & Results:

• Tabbed SQL editor for writing and managing multiple queries
• Syntax-highlighted editor with DuckDB SQL support
• Ctrl+R: Execute query
• Ctrl+N: Run query in a new results tab
• Ctrl+S: Save script to disk
• Ctrl+T: Create new editor tab
• Ctrl+W: Close current tab
• Results pane displays query output in scrollable tables

Right Pane — Profiler:

• Statistical summary for the selected source
• Row counts, column types, null rates, and value distributions
• Loaded on demand with P from the source pane

Explore Screen

Deep-dive analysis with statistical profiling and AI insights:

• Statistical Profile: Automatic data profiling showing distribution, nulls, and cardinality
• AI-Powered Insights: Ask questions about your dataset using LLM integration
• Interactive Q&A: Type questions to get natural-language analysis and recommendations
• Keyboard Shortcuts:
  • Ctrl+R or Enter: Submit question to AI
  • Q: Return to project screen

Navigation

• Tab / Shift+Tab: Move focus between panes
• Modal dialogs for source attachment and confirmations
• Themed UI with multiple color schemes (configurable via profile settings)

---

Changelog

v0.9.3 (2026-05-31)

Fix

• context not available in tui workers

Refactor

• core: simplify project loading

v0.9.2 (2026-05-31)

Refactor

• cli: identify sources by URI scheme, replace --path with --uri
• core: introduce Source protocol and URI-based data sources

v0.9.1 (2026-05-05)

Fix

• core: validate source name and path in source_attach

v0.9.0 (2026-05-03)

Feat

• core: introduce global settings.toml for app-level configuration

v0.8.2 (2026-05-03)

Fix

• tui: show column types in expanded source items
• tui: restore h/l source navigation when a source is expanded

v0.8.1 (2026-05-03)

Fix

• move dependencies back inside [project] table

v0.8.0 (2026-05-03)

Feat

• add navigation screen and demo project setup

Fix

• tui: make profiler on-demand via 'p' keybind
• core: skip configure_s3 for projects with no S3 sources
• tui: move NavigationScreen I/O off the main thread
• tui: use call_from_thread for notify calls inside thread workers

v0.7.1 (2026-04-27)

Fix

• toml properties not being overriden by corresponding env vars

v0.7.0 (2026-04-26)

Feat

• tui: support multiple tabs in editor and result panes

v0.6.0 (2026-04-26)

Feat

• core: read analyzer row limit setting
• tui: read theme setting from core section
• core: add core config section

v0.5.1 (2026-04-26)

Refactor

• tui: consolidate styles in app.tcss

v0.5.0 (2026-04-24)

Feat

• tui: add explore action to result pane

v0.4.0 (2026-04-22)

Feat

• data profile is presented on source selection
• enable ai powered data source exploration
• cli: add configuration profile delete and demo commands
• add anthropic section to configuration model

Fix

• analyzer mishandles bool types
• typo on app title

Refactor

• move data profiling and llm code to analyzer module
• move data profiling code from tui to core.analyzer
• split core.main responsibilities across submodules
• core

v0.3.4 (2026-04-20)

Fix

• tui: rebuild source list after attach and detach

v0.3.3 (2026-04-19)

Feat

• add commitizen for automated changelog and versioning

Fix

• Add a timeout for S3 source path existence check
• Project init ignores sources if unable to reach them or their existence check fails

v0.3.0 (2026-04-19)

Feat

• Add support for attaching S3 sources to a project

v0.2.2 (2026-04-18)

Feat

• Local sources can be attached to and detached from a project during runtime (#4)
• tui: basic source navigation and query execution
• core: can attach and query csv, parquet and parquet datasets (#3)
• tui: preliminary layout
• core: project context registers sources for querying (#2)
• wip: project lifecycle management (#1)

Refactor

• core,cli: moving around and renaming things