Developer SDKs
SDKs for
asset operations.
Pick your path: Python for automation, C++ for native tools, Dart for desktop apps. Same object model. Same events. Different runtimes.
Most devs ship their first integration in under 30 minutes.
Pick your path
Each SDK exposes the same object model. Choose based on your runtime.
Python SDK
For automation, pipelines, and AI tooling
C++ SDK
For performance-critical and native tools
Quickstart snippets
Copy, paste, ship. Each example shows a real operation.
Python — Create revision with lineage
pip install kumihoimport kumiho
# Connect to Kumiho Cloud (uses cached credentials)
kumiho.connect()
# Create or get a project
project = kumiho.create_project("my-project", "My Project")
# Create an asset space
space = project.create_space("characters")
# Create an item and revision
item = space.create_item("hero", kind="image")
revision = item.create_revision(
metadata={"prompt": "hero portrait, epic lighting"}
)
# Attach a file artifact (no upload)
revision.create_artifact(
name="output",
location="file:///projects/hero/hero_v01.png"
)
print(f"Created: {revision.kref}")C++ — Resolve kref to local path
git clone github.com/kumihoclouds/kumiho-cpp#include <kumiho/client.hpp>
auto client = kumiho::Client::create();
// Resolve kref to local file path
auto artifact = client->resolve_artifact(
"kref://project/characters/hero.image?r=latest&a=main"
);
std::cout << "Path: " << artifact.local_path << std::endl;Dart — Search items by metadata
dart pub add kumihoimport 'package:kumiho/kumiho.dart';
final client = KumihoClient();
// Search for items by metadata
final results = await client.searchItems(
query: 'hero',
kinds: ['image', 'texture'],
tags: ['approved'],
);
for (final item in results) {
print('${item.name}: ${item.kref}');
}What you don't have to build
Use Kumiho as your asset intelligence layer — not another weekend project.
Multi-tenant asset databases
Revision and lineage schemas
Event streaming infrastructure
Permission and access control
Full-text and metadata search
Graph queries and traversal
Estimated time saved: 3–6 months of backend work.
Ready to build?
Start with the SDK that fits your stack. Scale to events, automation, and agents when you need them.