Warehouse UI

Universal database IDE CLI — query PostgreSQL, MySQL, SQLite, BigQuery, MongoDB with cost projection

Security Scan

The skill's requirements and runtime instructions are coherent with a database IDE CLI: it expects a local 'warehouse-ui' binary, a DATABASE_URL for auto-connect, and optional AI provider keys for its 'ai' feature.

Purpose & Capability

Name/description (universal DB CLI) aligns with required binary 'warehouse-ui' and primary credential DATABASE_URL. Supported DBs (Postgres, MySQL, SQLite, BigQuery, MongoDB) and BigQuery service-account usage are consistent with the stated capabilities.

Instruction Scope

SKILL.md instructs the agent to run the warehouse-ui binary and to read DB connection URLs, local SQLite files, and (optionally) a BigQuery service-account JSON path; these are expected. It also documents using OPENAI_API_KEY or ANTHROPIC_API_KEY for AI features — this implies sending queries or prompts to external model APIs, which is expected for the feature but means query text or schema might be transmitted externally. The SKILL.md does not instruct reading unrelated system files.

Install Mechanism

SKILL.md and its embedded metadata recommend downloading releases from GitHub Releases (reasonable, low-risk). There is a minor registry inconsistency: the top-level registry metadata indicated 'No install spec' while SKILL.md includes a 'github-release' install entry. The install source (GitHub Releases) is acceptable but users should verify release authenticity (checksums/signatures) before running a shipped binary.

Credentials

PrimaryEnv DATABASE_URL is appropriate. The AI provider keys (OPENAI_API_KEY, ANTHROPIC_API_KEY) are referenced as optional for the 'ai' command but are not listed as required in the skill's top-level required-env metadata — this is reasonable for optional features. BigQuery's service-account JSON path is a necessary credential for that integration. Be mindful that using AI features will transmit query text (potentially sensitive) to third-party model APIs.

Persistence & Privilege

Skill is not always-enabled and does not request elevated platform privileges. It does not instruct modifying other skills or system-wide agent settings. Autonomous invocation is allowed (platform default) but not combined with any other high-risk setting.

Assessment

This skill appears to do what it says, but it runs a local binary you will need to install—verify the GitHub release (checksum/signature) before running. DATABASE_URL (or explicit connect) gives the tool DB access; prefer a least-privilege, read-only DB user for exploration. If you use BigQuery, provide a service-account JSON with only the needed permissions. The AI feature will send prompts/possibly schema or queries to OpenAI/Anthropic—avoid sending sensitive data there. If you are uncertain, review the upstream repository/releases and run the binary in an isolated environment (container or VM) before using it against production data.

Warehouse UI — Database Query Tool

Use this skill to connect to databases, explore schemas, run queries, estimate costs, and generate SQL from natural language.

Installation

Download from GitHub Releases: https://github.com/olegnazarov23/warehouse-ui/releases

• macOS: Download the DMG, drag to Applications, then add to PATH: ln -s /Applications/warehouse-ui.app/Contents/MacOS/warehouse-ui /usr/local/bin/warehouse-ui
• Windows: Run the installer EXE, it adds to PATH automatically

Supported Databases

• PostgreSQL
• MySQL
• SQLite
• BigQuery (with cost projection)
• MongoDB

Connect to a Database

Before running queries, establish a connection:

# From a connection URL
warehouse-ui connect --url "postgres://user:pass@localhost:5432/mydb"

# With explicit parameters
warehouse-ui connect --type postgres --host localhost:5432 --database mydb --user admin --password secret

# SQLite (local file)
warehouse-ui connect --type sqlite --database /path/to/data.db

# BigQuery (service account)
warehouse-ui connect --type bigquery --database my-gcp-project --option sa_json_path=/path/to/sa.json

# MySQL
warehouse-ui connect --url "mysql://user:pass@localhost:3306/mydb"

Check Connection Status

warehouse-ui status

Explore Schema

# List all databases
warehouse-ui schema list-databases

# List tables in a database
warehouse-ui schema list-tables --database mydb

# Describe a table (columns, types, nullability)
warehouse-ui schema describe users --database mydb

Run Queries

# SQL as argument
warehouse-ui query "SELECT * FROM users LIMIT 10"

# With explicit limit
warehouse-ui query --sql "SELECT count(*) FROM orders WHERE created_at > '2024-01-01'" --limit 1000

# From a SQL file
warehouse-ui query --file path/to/report.sql

Output is JSON with columns, rows, row count, duration, and (for BigQuery) bytes processed and cost.

Cost Estimation (Dry Run)

Check query cost before executing — especially useful for BigQuery:

warehouse-ui dry-run "SELECT * FROM big_dataset.events WHERE date > '2024-01-01'"

Returns: estimated bytes, estimated cost (USD), statement type, referenced tables, and warnings.

AI-Powered Queries

Generate SQL from natural language using a configured AI provider (set OPENAI_API_KEY or ANTHROPIC_API_KEY):

# Generate SQL only
warehouse-ui ai "show me the top 10 customers by total revenue"

# Generate and execute
warehouse-ui ai "find all orders from last week that were cancelled" --execute

List Saved Connections

warehouse-ui connections

Query History

warehouse-ui history --limit 10
warehouse-ui history --search "SELECT"

Disconnect

warehouse-ui disconnect

Output Format

All commands output JSON to stdout by default. Add --format table for human-readable output. Errors are JSON on stderr with exit code 1.

Environment Variables

• DATABASE_URL — Auto-connect without explicit connect step (supports postgres://, mysql://, sqlite://, mongodb://)
• OPENAI_API_KEY — Required for ai command with OpenAI
• ANTHROPIC_API_KEY — Required for ai command with Anthropic

Tips

• Set DATABASE_URL to skip the connect step entirely
• Use schema describe <table> to understand table structure before querying
• Use dry-run on BigQuery to check costs before executing expensive queries
• Use --limit to control result size for large tables
• Use connections to see databases already configured in the desktop app