Public maps · Bank of Anthos

Bank of Anthos, mapped from the code.

Google's sample bank: a Python Flask front end and account services, Java ledger services and two Postgres databases. Kuuhaku read its source at commit db35fea, with no configuration and no hand-fixing, and found 8 services in Python and Java, and 9 connections: 6 over HTTP and 3 through shared databases. Each one below says how it was found.

Open the live map Map your own repos

Services

Every repo or deployable the scanner found, with how many connections leave it and reach it.

Service	Stack	Role	Calls	Called by
`frontend`	Python · Flask	service	5	0
`balancereader`	Java · Spring Boot	service	0	4
`ledgerwriter`	Java · Spring Boot	service	2	1
`contacts`	Python · Flask	service	0	2
`transactionhistory`	Java · Spring Boot	service	1	1
`userservice`	Python · Flask	service	1	1
`ledgermonolith`	Java · Spring Boot	service	0	0
`loadgenerator`	Python	consumer	0	0

Connections

config: the address is written in configuration (an env file, a compose file, a proxy route) and the call is matched to it. inferred: the call's path matches a route of the target and nothing contradicts it. Call sites count the places in the code that make the call.

From	To	Type	Evidence	Call sites	How it was found
`frontend`	`balancereader`	HTTP	config	1	URL worked out at frontend.py:118
`frontend`	`contacts`	HTTP	config	2	URL worked out at frontend.py:130
`frontend`	`ledgerwriter`	HTTP	config	1	URL worked out at frontend.py:334
`frontend`	`transactionhistory`	HTTP	config	1	URL worked out at frontend.py:124
`frontend`	`userservice`	HTTP	config	2	URL worked out at frontend.py:445
`ledgerwriter`	`balancereader`	HTTP	config	1	URL worked out at src/main/java/anthos/samples/bankofanthos/ledgerwriter/LedgerWriterController.java:215
`ledgerwriter`	`balancereader`	shared database	config	—	SPRING_DATASOURCE_URL in ledgerwriter and SPRING_DATASOURCE_URL in balancereader name the same Postgres ledger-db/postgresdb
`transactionhistory`	`balancereader`	shared database	config	—	SPRING_DATASOURCE_URL in transactionhistory and SPRING_DATASOURCE_URL in balancereader name the same Postgres ledger-db/postgresdb
`userservice`	`contacts`	shared database	config	—	ACCOUNTS_DB_URI in userservice and ACCOUNTS_DB_URI in contacts name the same Postgres accounts-db/accounts-db

Left for a person

3 items the scanner would not settle on its own and put in the review inbox instead of guessing:

requests in frontend: base URL assigned at runtime (frontend.py:541)
metadata_url in frontend: base URL assigned at runtime (frontend.py:691)
client in loadgenerator: base URL assigned at runtime (locustfile.py:49)

Other systems

OpenTelemetry Demo: 23 services, 30 connections
Train Ticket: 48 services, 140 connections
Online Boutique: 12 services, 16 connections
eShop: 19 services, 31 connections
Robot Shop: 9 services, 12 connections
Sock Shop: 8 services, 7 connections
Quarkus Super Heroes: 7 services, 6 connections
Spring PetClinic Microservices: 7 services, 14 connections

Your own system

Run npx kuuhaku scan . in a folder of repos: it reads the code on your machine and only the graph leaves it. Or connect GitHub or Bitbucket and let the hosted runner scan with a read-only key. The map, a page per repo, a review inbox and a check on every pull request follow. Start free.