See how a codebase is put together.

Point Pocket Blueprint at a git repository, an archive or a folder and get its structure, a graph of what uses what, every HTTP endpoint with an OpenAPI document, the database tables from the migrations and the environment variables. Nothing from the code is ever run.

Get it on GitHub How to run it

Included in Pocket Box
Sign in with Pocket Core

A made-up shop's codebase on an iPhone: files, lines, endpoints and tables, its languages and largest folders

Graph

Which folder uses which.

Folders or files as circles sized by lines, arrows for imports, thicker for more of them. Folders that import each other are drawn in red, so a loop is easy to spot. Double-click to go inside.

The folder graph of a shop: the API uses config, mail, store and billing, and store and billing import each other in red

API

Every endpoint, with a ready OpenAPI file.

Routes are read from the code and from API documents, grouped by path or by file with the handler and the line. A note says which endpoints are in the code but missing from the document.

chi, gin, echo, Express, FastAPI, Django, Spring, Rails and more
Download an OpenAPI document made from what the code serves
Filter by method, path, handler or file

The endpoints of a shop grouped by path, each with its method, handler and file line

Data

The tables, straight from the migrations.

SQL migrations are read in order, so each table shows its columns, types, keys and constraints as they end up, with the file that made it and the tables that refer to it.

Prisma schemas too
A diagram of the references between tables
Environment variables, and which ones nobody documented

Six tables of a shop with columns and types: coupons, customers, order items and orders with their references

Source

Open any file at the line.

Jump from a function, an endpoint or a search result to the source, opened at the right line. Each file shows who uses it, its functions and types, and the notes you keep on it.

For Go, who calls what, read with the Go parser itself
Hotspots: big files that change often, from the git history
Ask the AI to explain a folder or a file

A billing file opened at line 34 on an iPhone, the charge and refund functions around it

And the rest of it

Compare two scans
See what changed between scans: new endpoints, tables, files, packages and variables. Scan again every 6 hours, daily, weekly or from CI.
Code from anywhere
A git address with a branch and a read-only token, a zip or tar archive, a folder in Pocket Files, or a folder on the server that an admin allows.
Packages and services
Dependencies from go.mod, package.json, pyproject and more, with the files that really import them, and the services from docker compose.
Read, never run
Go is read exactly; other languages with careful patterns. Server folders can only be scanned inside the roots you list.

Install

One container, one volume.

Run the image and sign in. To scan folders on the server, mount them read-only and list them in SOURCE_ROOTS; git repositories and archives need nothing extra.

On a Pocket Box it is already installed and can scan folders from Pocket Files.

Every setting in the README →

docker run -d --name pocket-blueprint -p 8080:8080 -v ./data:/data \
  -e ADMIN_EMAIL=you@example.com -e ADMIN_PASSWORD=change-me-please \
  ghcr.io/abjelosevic88/pocket-homelab-blueprint:latest

See how a codebase is put together.

Which folder uses which.

Every endpoint, with a ready OpenAPI file.

The tables, straight from the migrations.

Open any file at the line.

And the rest of it

Compare two scans

Code from anywhere

Packages and services

Read, never run

One container, one volume.

More in Work and code