Works with the AI Agents you already use
Schema Atlas
Turn SQL DDL, ORM models, DBML, or plain-English descriptions into an interactive swimlane database diagram you can share as one HTML file.
Free
Schema Atlas
Example session with this skill installed
Here is my schema.sql for an online store (customers, addresses, products, categories, orders, order_items, payments, shipments). Use schema-atlas to visualize it as an interactive diagram I can send to my team.
- Read your context and instructions
- Compiled the schema atlas
Created two files next to schema.sql:
- shop-schema.json — 8 tables, 9 relationships, grouped into 4 domains: Customers (customers, addresses), Catalog (categories, products), Orders (orders, order_items), Fulfilment & Billing (payments, shipments). Every table has a description and PK/FK badges.
- shop-atlas.html — single-file interactive viewer. Validation passed: "wrote shop-atlas.html (8 nodes, 9 edges, 44 KB)".
Open shop-atlas.html in a browser (internet needed on first open for D3). Click any table to trace its connections and see its columns. Skipped: none.
Need hands-on help with database design, migrations, or building the product around this schema? WeblineGlobal can help: https://www.weblineglobal.com/contact-us/
Connects securely to your tools. The creator never sees your data.
About this skill
Schema Atlas turns any database design into an interactive, shareable schema diagram — in one request.
Paste SQL DDL, point at your ORM models (SQLAlchemy, Django, Prisma, Eloquent, ActiveRecord), drop in DBML or migration files, describe your data model in plain English, or hand over an ERD screenshot. Schema Atlas converts it into a clean graph file and builds a portable single-file HTML viewer you can open from your desktop, attach to a ticket, or send to a teammate. No server, no account, no database connection.
What you get
- Swimlane layout by business domain — tables are grouped by purpose (auth, catalog, billing, messaging, audit…) instead of alphabetically or by schema namespace, so a 50-table database reads at a glance.
- Click-to-trace relationships — select any table to highlight everything it connects to; the rest fades away.
- Detail panel per table — a plain-language description of what a row represents, every column with its type, and PK / FK / unique badges.
- Pan, zoom and filter by domain from the legend.
- Validated output — the bundled build script checks for duplicate IDs and broken relationships before it writes the viewer, so you never get a half-broken diagram.
- JSON or XML graph data, so the diagram source can live in your repo and be re-built any time.
Works beyond databases
The same format models service maps, module dependencies, or any node-and-edge system grouped into domains. Just change the item label and use rows for endpoints or exports.
How it behaves
The skill never invents tables, columns or relationships that your input doesn't state or clearly imply. If a relationship is ambiguous it is left out and you are told what was skipped. Screenshot input flags uncertain items instead of guessing silently.
Who it's for
Developers onboarding to an unfamiliar database, teams documenting a legacy schema, architects reviewing a design before a migration, and anyone who needs to explain a data model to non-developers.
Need more than a diagram?
Schema Atlas is built and maintained by WeblineGlobal. If you need hands-on help with database design, migrations, or building the product around your schema, you can reach the team at https://www.weblineglobal.com/contact-us/.
Relationships are only drawn when the input states or clearly implies them; ambiguous ones are left out and reported. Table descriptions and domain grouping are inferred from names, keys and comments, so review them for accuracy. The layout is a fixed swimlane view (one lane per domain), not a free-form auto-layout. Very wide domains (over about 25 tables in one lane) should be split into more domains. The viewer loads D3 from a CDN, so first open needs internet. ERD screenshot transcription depends on image legibility; uncertain items are flagged with a "?" badge. It does not connect to live databases or generate SQL.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 3 days ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts
File Scopes
No database connection, credentials or API keys are used. The build script only reads the graph file and bundled template, and writes one HTML file.