Open source — MIT license

arc-tables

Point it at any Postgres or Supabase database.
Get beautiful interactive HTML maps and a plain-English audit with the exact SQL to fix every issue.

Get started View on GitHub

arc-tables — zsh

$ arc-tables connect supabase --project pqwfnhblts --as arc-prod

✔ Connected to arc-prod (166 tables, 12 views)

$ arc-tables audit --profile arc-prod --format cards --output audit.html

Introspecting schema public...

Running 9 audit rules across 166 tables...

✔ Found 144 issues — 12 errors · 67 warnings · 65 info

✔ Wrote card report: audit.html

✔ Wrote SQL: fixes.sql

$ open audit.html

Schema visualization

Your database at a glance

Every table, every foreign key, color-coded by group. Hover to isolate. Click to explore. All in a single HTML file — no server, no deps.

companies

iduuid

company_nametext

domaintext

contacts

iduuid

full_nametext

emailtext

leads

iduuid

company_iduuid

stagetext

clients

iduuid

company_iduuid

lead_iduuid

created_atts

activebool

client_contacts

client_iduuid

contact_iduuid

client_transactions

iduuid

client_iduuid

amountint

stripe_clients

client_iduuid

stripe_idtext

 FK link
      

 PK
        
 FK
        
 text
        
 int

🗺

Full ecosystem map

Every table laid out with color-coded groups, animated FK flow lines, and hover-to-isolate. Single HTML file, no server needed.

📊

Lifecycle flow diagrams

Trace how data moves from leads through to clients and transactions. Pick any path through your schema and get a focused flow view.

🧾

Plain-English audit cards

Tinder-style issue review. Each card explains the problem, why it matters, and the fix in plain English — not just the raw SQL.

🔗

Diagram per issue

Every audit card has a diagram button that centers the affected table on an interactive canvas with all FK neighbors. Pan & zoom to explore.

Audit review

Issues, not noise

One card at a time. Accept the fix, skip it, or dismiss. Notes persist to localStorage. Each card links to a live diagram of the affected table.

connection

⚠ warning

client_information . client_name

This column uses a name to link tables instead of an ID

What: client_information.client_name references clients.name as text instead of using clients.id.

Why it matters: If a client is ever renamed, this link silently breaks. Name-based joins are also slower and can't enforce referential integrity.

Fix: Add a client_id uuid column with a real FK constraint, backfill from the name match, then drop the old text column.

✗ Dismiss

Skip

✓ Accept fix

issue 7 of 144 · 3 accepted · 1 dismissed

Three commands, full picture

Connect

Point arc-tables at your Supabase project or Postgres connection string. Credentials stay on your machine.

Introspect

Reads all tables, columns, foreign keys, views, and indexes. No schema changes, read-only queries only.

Generate

Outputs a self-contained HTML file. Map, flow diagram, or card-based audit report — your choice.

Fix

Review the issues in your browser. Accept fixes to build a fixes.sql file. Run it when you're ready.

Built-in audit rules

9 rules, zero config

Every rule ships with the exact SQL to fix the problem and plain-English explanations anyone can read.

Duplicate-prefix orphansFinds plural/singular mismatches like clients_X vs client_X

String FK columnsDetects name-based links that should be ID-based FK constraints

Ambiguous name columnsFlags bare name columns that collide across joins

Deprecated columnsFinds *_deprecated columns left in place after migration

Orphan tablesTables with no FK in or out — likely abandoned or missing a link

Single-row satellitesTables that could be columns — one row per parent, no time axis

Empty staging tablesFinds *_staging and temp_* tables with zero rows

Casing duplicatesDetects Foo and foo both existing in the same schema

Broken viewsViews that reference columns or tables that no longer exist

Installation

Get started in 30 seconds

Install globally via npm, connect to your database, and generate your first audit.

Install

# install globally

$ npm install -g @arc-web/arc-tables

# or run without installing

$ npx @arc-web/arc-tables --help

Quick start

# connect to Supabase

$ arc-tables connect supabase --project your-project-ref --as mydb

# generate the full schema map

$ arc-tables map --profile mydb --output schema.html

# run the audit (card view)

$ arc-tables audit --profile mydb --format cards --output audit.html

✔ Open audit.html in your browser. Review issues, accept fixes, run fixes.sql.