Isis is Oshun's generative AI factory, responsible for all AI-powered content generation including images, video, audio, 3D models, and text.
Overview#
| Aspect | Description |
|---|---|
| Domain | Generative AI & Asset Creation |
| Scope Tag | scope:isis |
| API Port | 4001 |
| Database Schema | isis |
Capabilities#
- Image Generation: Stable Diffusion, DALL-E, Midjourney-style
- Video Generation: Zeroscope, Runway, text-to-video
- Audio Generation: TTS, music, sound effects
- 3D Generation: Point-E, NeRF, Gaussian splatting
- Text Generation: LLM integration (GPT-4, Claude, Gemini)
- Workflow Orchestration: Multi-step AI pipelines
Libraries#
text
libs/isis/
├── client/ # TypeScript client for other domains
├── job-envelope/ # Canonical queue envelope schema
├── database/ # Prisma schema & utilities
├── workflows/ # ComfyUI workflow registry
├── comfyui-nodes/ # Custom ComfyUI nodes
├── ai-providers/ # Provider adapters
│ ├── openai/
│ ├── anthropic/
│ ├── stability/
│ ├── elevenlabs/
│ ├── runway/
│ └── replicate/
├── 3d-generation/ # 3D asset generation
├── 3d-browser/ # Browser-first 3D viewing and editing
├── ai-video/ # Video generation
├── audio-generation/ # Audio/TTS generation
├── outputs/ # Output manifest management
└── gaussian-splatting/ # NeRF/Gaussian splatting
Quick Start#
Using the Client#
typescript
import { createIsisClient } from '@isis/client';
const isis = createIsisClient({
baseUrl: process.env.ISIS_API_URL || 'http://localhost:4001',
});
// Generate an image
const job = await isis.generation.create({
type: 'image',
prompt: 'A majestic dragon flying over mountains',
model: 'stable-diffusion-xl',
parameters: {
width: 1024,
height: 1024,
steps: 50,
},
});
// Wait for completion
const result = await isis.generation.waitForCompletion(job.id);
console.log(`Generated: ${result.url}`);
Running Locally#
bash
# Start Isis API
nx serve isis-api
# Run tests
nx test @isis/client
# Build all Isis libraries
nx run-many --target=build --projects=tag:scope:isis
API Endpoints#
| Method | Endpoint | Description |
|---|---|---|
| POST | /v1/generation/jobs |
Create generation job |
| GET | /v1/generation/jobs/:id |
Get job status |
| GET | /v1/generation/jobs/:id/result |
Get generation result |
| DELETE | /v1/generation/jobs/:id |
Cancel job |
| GET | /v1/assets/:id |
Get generated asset |
| GET | /v1/models |
List available models |
| POST | /v1/workflows |
Execute workflow |
Events#
| Event | Description |
|---|---|
isis.job.started |
Generation job started |
isis.job.progress |
Job progress update |
isis.job.completed |
Job completed successfully |
isis.job.failed |
Job failed |
isis.asset.generated |
Asset created from job |
Configuration#
bash
# Required
ISIS_DATABASE_URL=postgresql://...?schema=isis
REDIS_URL=redis://localhost:6379
# AI Providers
ISIS_OPENAI_API_KEY=sk-...
ISIS_ANTHROPIC_API_KEY=sk-ant-...
ISIS_REPLICATE_API_TOKEN=r8_...
ISIS_STABILITY_API_KEY=sk-...
# Storage
ASSET_BUCKET=oshun-assets
ASSET_CDN_URL=https://cdn.oshun.io
Local Development#
Start the full development environment:
bash
# Start infrastructure (PostgreSQL, Redis, MinIO)
docker compose -f docker/docker-compose.dev.yml up -d
# Run database migrations
pnpm nx run @isis/database:prisma:migrate:dev
pnpm nx run @isis/workflow-registry:db:migrate
# Start the Generation API
pnpm nx serve isis-generation-api
Note: GPU workers require NVIDIA GPU with CUDA support. See Docker profile
with-gpu.
See CLAUDE.md for detailed local development setup.