whisper-money/.agents/skills/querying-the-database/SKILL.md

name	description	metadata
querying-the-database	Query the local or production database from the CLI via the `agent:db` artisan command. Activates when the user asks to inspect, count, look up, or run a query against the database; asks 'how many X', 'what's in prod', 'check the prod DB', 'query the database'; or mentions production data, the prod database, or running SQL.	author whisper-money

Querying the Database

When to Apply

Activate this skill whenever you need to read data from the database to answer a question — especially anything about production data. Prefer this command over the tinker, database-query, or database-schema Boost tools when the user asks about prod, since those default to the local connection.

The agent:db command

Runs a read query (SELECT, SHOW, DESCRIBE, etc.) and prints the result.

php artisan agent:db "<query>"

Options

• --format=json (default) — pretty-printed JSON, best for parsing the result yourself.
• --format=table — classic console table, best when showing the result to the user.
• --prod — run against the production database (the prod connection backed by PROD_DB_URL). Without this flag the query runs against the local DB.

Examples

# Local, JSON (default)
php artisan agent:db "select id, email from users limit 5"

# Local, human-readable table
php artisan agent:db --format=table "select count(*) as total from transactions"

# Production
php artisan agent:db --prod "select count(*) from users"
php artisan agent:db --prod --format=table "select status, count(*) from subscriptions group by status"

Guidelines

• Read-only: the command uses DB::select(), so only read statements work. It will not run INSERT/UPDATE/DELETE. Never attempt to mutate prod data this way.
• Be careful with --prod: this is live customer data. Only run prod queries the user explicitly asked for, keep them scoped (add LIMIT, filter by id), and never dump large or sensitive datasets unprompted. This app is privacy-first.
• Use --format=json when you need to read the values to continue working; use --format=table when presenting results back to the user.
• Inspect schema first with database-schema (local) when you're unsure of column names before writing a query.