██╗  ██╗ █████╗ ██╗   ██╗███████╗██████╗  ██████╗  ███████╗████████╗██╗   ██╗██████╗ ██╗ ██████╗
██║ ██╔╝██╔══██╗██║   ██║██╔════╝██╔══██╗██╔═══██╗ ██╔════╝╚══██╔══╝██║   ██║██╔══██╗██║██╔═══██╗
█████╔╝ ███████║██║   ██║█████╗  ██████╔╝██║   ██║ ███████╗   ██║   ██║   ██║██║  ██║██║██║   ██║
██╔═██╗ ██╔══██║╚██╗ ██╔╝██╔══╝  ██╔══██╗██║   ██║ ╚════██║   ██║   ██║   ██║██║  ██║██║██║   ██║
██║  ██╗██║  ██║ ╚████╔╝ ███████╗██║  ██║╚██████╔╝ ███████║   ██║   ╚██████╔╝██████╔╝██║╚██████╔╝
╚═╝  ╚═╝╚═╝  ╚═╝  ╚═══╝  ╚══════╝╚═╝  ╚═╝ ╚═════╝  ╚══════╝   ╚═╝    ╚═════╝ ╚═════╝ ╚═╝ ╚═════╝

The open-source Canva alternative for AI design.

bring your own models · own your files · canvas AI · self-hostable · local-first

The open design workspace for the AI era.

Canva made design easy. Kavero makes it yours.

Kavero is an open-source design workspace for people who want AI-powered creative tools without being locked inside one closed platform. It gives you a familiar visual canvas, generated asset history, AI-assisted editing workflows, provider-key control, and a local-first path you can run on your own machine.

Why Kavero

Creative tools are turning into closed bundles: their models, their storage, their workflow, their pricing. Kavero is built around a different idea.

You should be able to bring your own models, keep control of your files, run the core product yourself, and still get a polished design workspace that feels approachable from the first click.

Kavero is for creators, founders, marketers, small agencies, and builders who need to make social posts, launch graphics, ads, thumbnails, pitch visuals, and client assets without handing the whole workflow to a closed suite.

Highlights

• Familiar canvas: create designs with text, shapes, images, pages, assets, and PNG export.
• AI generation: generate visuals with prompt presets, references, model settings, and saved history.
• Canvas AI: use AI workflows for layout help, segmentation, asset generation, and image judging.
• Gallery: keep generated images organized and bring them back into your work.
• BYOK provider keys: keep AI access tied to your own provider accounts.
• Own your storage: use local managed storage or connect external storage flows.
• Local-first setup: run Kavero with the bundled Docker Compose Supabase stack.
• Self-hostable foundation: configure Kavero against your own Supabase-backed deployment.

Get Started

Prerequisites:

• Node.js 22+ or 24+
• pnpm
• Docker Desktop or Docker Engine

pnpm install
pnpm setup
pnpm setup:run

Open Kavero:

http://127.0.0.1:3000

The setup wizard creates local env files safely, protects existing secrets, and can start the local Docker stack for you.

Choose Your Setup

Local Docker

Recommended for trying Kavero with the full local stack.

pnpm setup
pnpm setup:run

Local Docker runs:

• Kavero app
• LiteLLM gateway, internal to the Docker network
• Supabase Auth
• Postgres
• PostgREST
• Supabase Storage
• Supabase bootstrap checks
• Local managed file storage through a Docker volume

Local Docker uses email/password sign-in and this storage contract:

KAVERO_DEPLOYMENT_PROFILE=local-first
KAVERO_AUTH_MODE=password
KAVERO_STORAGE_PROVIDER=kavero-managed
KAVERO_MANAGED_STORAGE_BACKEND=local-filesystem
KAVERO_LOCAL_STORAGE_ROOT=/data/kavero-storage

/data/kavero-storage is a container path backed by the kavero-local-storage Docker volume.

Manual Docker command:

docker compose --env-file .env.docker.local up --build

The LiteLLM service is not published to the host by default. Kavero receives only server-side gateway envs inside the Docker network.

pnpm setup generates the local LiteLLM secrets and can collect optional server-side Gemini, OpenAI, Groq, and Ollama values for the gateway. Normal setup does not require editing docker/litellm/config.yaml.

Host Development

Use this path when you want to run the Next.js dev server on your host machine while keeping LiteLLM in Docker as a local dependency.

cp .env.local.example .env.local
pnpm gateway:up
pnpm dev

On Windows PowerShell:

Copy-Item .env.local.example .env.local
pnpm gateway:up
pnpm dev

Host development uses:

• .env.local for the Next.js app running on your machine
• compose.dev.yml to publish only LiteLLM on 127.0.0.1:4000
• compose.yml for the same LiteLLM config used by Local Docker

Set these in .env.local:

NEXT_PUBLIC_SITE_URL=http://localhost:3000
KAVERO_API_ORIGIN=http://localhost:3000
KAVERO_MODEL_GATEWAY=litellm
KAVERO_LITELLM_BASE_URL=http://127.0.0.1:4000
KAVERO_LITELLM_API_KEY=sk-change-me
LITELLM_MASTER_KEY=sk-change-me
GEMINI_API_KEY=

For the current local dev gateway, KAVERO_LITELLM_API_KEY should match LITELLM_MASTER_KEY. Add a real GEMINI_API_KEY or another provider key for live model calls.

If you use Google auth while developing on localhost:3000, add this redirect URL in Supabase Auth:

http://localhost:3000/auth/callback

Stop the dev gateway when finished:

pnpm gateway:down

Cloud / Self-Host

Use this path when you already have a Supabase project or a hosted Supabase-compatible stack.

pnpm setup
pnpm kavero doctor cloud-self-host
pnpm build
pnpm start

Minimum env shape:

NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY=
SUPABASE_SERVICE_ROLE_KEY=
NEXT_PUBLIC_SITE_URL=
KAVERO_DEPLOYMENT_PROFILE=cloud
KAVERO_AUTH_MODE=google

Optional provider-owned storage:

GOOGLE_DRIVE_CLIENT_ID=
GOOGLE_DRIVE_CLIENT_SECRET=

Optional Kavero-managed storage:

KAVERO_STORAGE_PROVIDER=kavero-managed
KAVERO_MANAGED_STORAGE_BACKEND=supabase-storage

Optional hosted LiteLLM gateway:

KAVERO_MODEL_GATEWAY=litellm
KAVERO_LITELLM_BASE_URL=
KAVERO_LITELLM_API_KEY=
KAVERO_LITELLM_ROUTING_SECRET=

Keep LiteLLM credentials, the routing secret, upstream provider keys, and internal gateway URLs server-only. A hosted external LiteLLM service must run the matching Kavero custom-auth hook with the same routing secret and pinned LiteLLM contract; an unmodified LiteLLM endpoint is not supported for dynamic client routing.

Azure OpenAI orchestration and image generation use independent logical configurations. Orchestration uses the AZURE_API_* record; Azure GPT Image 2 uses the AZURE_IMAGE_* record. The same resource URL and key may be entered explicitly in both records, but values are never copied or inferred between them. See Azure OpenAI setup.

OpenAI orchestration offers stable Kavero aliases for gpt-5.6, gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna.

First Run

1. Open http://127.0.0.1:3000.
2. Create an account.
3. Go to Settings -> API keys.
4. Add your AI provider key.
5. Open /generate or /canvas.

From there, generate assets, keep them in Gallery, place them into the canvas, edit the design, and export when ready.

Configuration

File	Purpose
.env.docker.local	Local Docker stack. Created by pnpm setup.
.env.local	Local development or Cloud/self-host env.
.env.local.example	Host development reference for pnpm dev plus Docker LiteLLM.
.env.example	Manual Cloud/self-host reference.
.env.docker.local.example	Manual Local Docker reference.
compose.dev.yml	Host development override that publishes LiteLLM to localhost.

Important envs:

Env	Used for
NEXT_PUBLIC_SUPABASE_URL	Browser-facing Supabase URL.
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY	Browser/server Supabase client key.
SUPABASE_SERVICE_ROLE_KEY	Admin-backed app flows and provider-key storage.
NEXT_PUBLIC_SITE_URL	Public Kavero app origin.
KAVERO_DEPLOYMENT_PROFILE	cloud or local-first.
KAVERO_AUTH_MODE	google, password, or google-password.
KAVERO_STORAGE_PROVIDER	Selects new-write storage behavior.
KAVERO_MANAGED_STORAGE_BACKEND	supabase-storage or local-filesystem.
KAVERO_LOCAL_STORAGE_ROOT	Absolute storage root, or /data/kavero-storage in Docker.
KAVERO_MODEL_GATEWAY	Server-side model gateway selector for LiteLLM-backed model calls.
KAVERO_LITELLM_BASE_URL	Server-side LiteLLM base URL. Uses http://litellm:4000 in Docker and http://127.0.0.1:4000 for host development.
KAVERO_LITELLM_API_KEY	Server-side Kavero-to-LiteLLM credential.
KAVERO_LITELLM_ROUTING_SECRET	Separate server-only HMAC secret shared with the matching Kavero LiteLLM custom-auth hook.
KAVERO_MODEL_GATEWAY_CREDENTIAL_MODE	Server-side credential policy: env-or-user (default), user-required, or env-only.
LITELLM_MASTER_KEY	Local LiteLLM service master key. Generated by setup for Local Docker.
GEMINI_API_KEY	Optional server-side Gemini key for the LiteLLM gateway.
OPENAI_API_KEY	Optional server-side OpenAI key for the LiteLLM gateway.
GROQ_API_KEY	Optional server-side Groq key for the LiteLLM gateway.
AZURE_API_KEY	Optional server-side Azure OpenAI key; requires all other Azure values.
AZURE_API_BASE	Azure OpenAI HTTPS endpoint such as https://resource.openai.azure.com.
AZURE_API_VERSION	Azure OpenAI API version used for orchestration requests.
AZURE_DEPLOYMENT_NAME	Account-specific Azure OpenAI deployment name; kept server-only.
AZURE_BASE_MODEL	Curated routing family: gpt-4o, gpt-4.1, gpt-5, gpt-5.6-sol, gpt-5.6-terra, or gpt-5.6-luna. GPT-5 families use LiteLLM's explicit azure/gpt5_series/<deployment> route.
AZURE_IMAGE_API_KEY	Azure image-slot key, supplied independently even when identical to AZURE_API_KEY.
AZURE_IMAGE_API_BASE	Azure image-slot resource URL, supplied independently even when identical to AZURE_API_BASE.
AZURE_IMAGE_API_VERSION	Validated Azure GPT Image 2 API version: 2024-02-01.
AZURE_IMAGE_DEPLOYMENT_NAME	Azure GPT Image 2 deployment name; kept server-only.
AZURE_IMAGE_BASE_MODEL	Validated image family: gpt-image-2.
OLLAMA_BASE_URL	Optional server-side Ollama URL for the LiteLLM gateway.

The setup CLI keeps existing non-empty values by default, asks before replacing secrets, writes through a backup, and never echoes secrets back to your terminal.

Commands

Command	Description
pnpm setup	Guided setup wizard.
pnpm setup:doctor	Check Local Docker configuration.
pnpm kavero doctor cloud-self-host	Check Cloud/self-host configuration.
pnpm setup:run	Run Local Docker with docker compose up --build.
pnpm gateway:up	Start the LiteLLM Docker dependency for host development.
pnpm gateway:down	Stop the LiteLLM Docker dependency for host development.
pnpm dev	Start the Next.js development server.
pnpm build	Build for production.
pnpm start	Start the production server.
pnpm test	Run Vitest.
pnpm test:smoke	Run Playwright smoke tests.

Architecture

src/app                    Next.js routes and API route entrypoints
src/modules/canvas         Canvas editor runtime
src/modules/canvas/ai      Canvas AI workflows
src/modules/generation     Standalone /generate runtime
src/modules/gallery        Generated image history and Gallery runtime
src/modules/storage        Managed and connected storage foundation
src/modules/assets         Canvas asset helpers
src/modules/editor-panels  Canvas panel UI
src/lib                    Shared auth, Supabase, Drive, provider-key, and utility code
supabase                   Schema and local foundation SQL
docker                     Local Supabase Compose support
scripts                    Setup, bootstrap, and verification helpers

Development

pnpm install
cp .env.local.example .env.local
pnpm gateway:up
pnpm dev

pnpm dev runs the app on the host. If you want LiteLLM-backed AI features during host development, keep pnpm gateway:up running and point .env.local at http://127.0.0.1:4000.

Useful checks:

pnpm exec vitest run scripts/setup
pnpm exec vitest run src/modules/generation
pnpm exec tsc --noEmit
pnpm build

Local Supabase helper scripts are available when you want to work outside the Docker Compose app flow:

pnpm supabase:local:start
pnpm supabase:local:bootstrap
pnpm supabase:local:smoke

License

Kavero is released under the Apache License 2.0.