Akte — API

API Reference

REST API for programmatic access to OKTA Studio

Authentication

All authenticated endpoints require a Bearer token in the Authorization header. API keys start with onyx_sk_ and can be generated from your account settings.

bash

curl https://okta.studio/api/v1/credits \
  -H "Authorization: Bearer onyx_sk_your_api_key_here"

Keep your API key secret. Do not expose it in client-side code, public repositories or logs. Rotate keys immediately if compromised.

Base URL

https://okta.studio/api/v1

Agent Kit Download v0.4

Download the OKTA CLI and the matching coding-agent skill directly from the docs page. The kit ships with the BRIEF-DNA prompt composer, a live customer-facing pricing bridge, the Director cinematic vocabulary, ready-to-use anti-pattern and style libraries, and a hardened set of guardrails so agents render brand text and logos in-prompt, stay on the right model, and use all available image references instead of one.

New in v0.4: agent install commands now include the extracted okta-agent-kit folder · media-library upload, download and protected URL handoff are documented as the default agent workflow · product-photo to reference-image to video-clip automation has a canonical playbook · 90-second videos are treated as planned multi-clip jobs until a dedicated storyboard/render endpoint exists.

zip

Full Agent Kit

ZIP with CLI, shell wrappers, the complete skill bundle (SKILL.md, 11 references, 2 config files, 6 helper scripts), agent config, and LLM README. Single file, drop into any agent's skills directory.

mjs

CLI Script

Raw CLI executable. Supports --image-urls for up to 10 references (comma-separated or repeated). Use when your agent only needs the executable and handles placement itself.

Skill Entry Point

SKILL.md entry point for skill-aware runtimes. Use the full kit when the agent also needs the references, config and helper scripts it routes into.

mjs

Pricing Bridge

quote-price.mjs — live customer credit + euro quote derived from src/utils/pricing.ts. Drop into any skill or invoke directly.

Prompt Anatomy

prompt-anatomy.md — the 12-slot BRIEF-DNA composer with hardened Slot 6, brand-asset rendering rules, and the phase-style input mapping. Use as a standalone reference for the prompt-engineering layer.

yaml

Agent Config

openai.yaml metadata for skill pickers and default prompts in agent runtimes.

LLM README

Machine-oriented orientation for Codex, Claude Code, and similar coding agents.

cmd

Windows Wrapper

Direct .cmd launcher for PowerShell and Windows terminals.

POSIX Wrapper

Shell launcher for macOS and Linux environments.

PowerShell install

powershell

Invoke-WebRequest https://okta.studio/downloads/okta-agent-kit -OutFile okta-agent-kit.zip
Expand-Archive .\okta-agent-kit.zip -DestinationPath .
Set-Location .\okta-agent-kit
$env:OKTA_API_KEY = "onyx_sk_your_api_key_here"
.\okta.cmd --json models list --type image

macOS / Linux install

bash

curl -L https://okta.studio/downloads/okta-agent-kit -o okta-agent-kit.zip
unzip okta-agent-kit.zip
cd okta-agent-kit
export OKTA_API_KEY="onyx_sk_your_api_key_here"
sh ./okta --json models list --type image

The recommended flow is: create an API key in your account, download the full kit, point the coding agent atOKTA_LLM_README.md, then place theskills/okta-clifolder into your agent workspace if needed, then use the wrapper command for repeatable JSON-based API calls.

Agent Automation Playbook

Use this sequence when a coding agent should turn an existing product photo into reusable references and then into video clips. The media library is the handoff layer: uploaded files, generated outputs and selected references all become protected asset URLs that the API resolves server-side when the same Bearer token is used.

1. Upload or select the product anchor

powershell

.\okta.cmd --json assets upload --file ./input/product.png --name product.png --retention-type pinned
.\okta.cmd --json assets list --source upload --search product --limit 10
.\okta.cmd --json assets select --id <product_asset_id>

2. Generate reusable reference images

powershell

.\okta.cmd --json image edit `
  --model gpt-image-2 `
  --prompt "Create a clean ecommerce hero reference from this product photo; preserve product identity and visible text." `
  --image-url "https://okta.studio/api/v1/assets/<product_asset_id>/content" `
  --aspect-ratio 1:1 `
  --quality medium `
  --output ./output/reference-01.png

3. Persist generated references for the next agent step

powershell

.\okta.cmd --json assets upload --file ./output/reference-01.png --name reference-01.png --retention-type pinned
.\okta.cmd --json assets select --id <reference_asset_id> --output ./output/reference-01-check.png

4. Create video clips from selected references

powershell

.\okta.cmd --json video create `
  --model veo-3.1-fast `
  --prompt "Shot 01: slow cinematic orbit around the product, premium studio lighting, no text overlays." `
  --image-url "https://okta.studio/api/v1/assets/<reference_asset_id>/content" `
  --duration 8 `
  --aspect-ratio 16:9 `
  --wait `
  --output ./output/shot-01.mp4

Protected URLs: asset URLs are not public CDN links. Send the same API key or let the CLI download them with --output.

Long videos: current public video models create short clips. A 90-second result should be planned as multiple 6-15 second shots and assembled after generation.

Agent memory: save job IDs, prompts, asset IDs, output paths and credit usage so Codex or Claude Code can continue the campaign without guessing.

Available Image Models

Prices shown here are customer-facing defaults. Use GET /api/v1/models for the current live credit values.

Model ID	Name	Default Price	Edit	Resolutions
nano-banana-2	Nano Banana 2	16 credits		0.5K, 1K, 2K, 4K
nano-banana-lite	Nano Banana Lite	8 credits		auto
nano-banana-pro	Nano Banana Pro	39 credits		1K, 2K, 4K
gpt-image-2	GPT Image 2	29 credits default		1K, 2K, 4K
krea-v2-large	Krea 2 Large	12 credits	-	auto
ideogram-v4	Ideogram 4	12 credits	-	1K, 2K
grok-imagine-image	Grok Imagine	14 credits	-	1K, 2K
seedream-v45	Seedream 4.5	8 credits		1K, 2K, 4K
seedream-v5-pro	Seedream 5 Pro	13-26 credits		1K, 2K
seedream-v5-lite	Seedream 5 Lite	7 credits		2K, 3K, 4K

Endpoints

GET/api/v1/models

List Models

Retrieve all available image and video generation models with their capabilities and pricing.

Returns every model available for generation, grouped by type. Use the id field when calling the generate endpoints.

Example Request

bash

curl https://okta.studio/api/v1/models

Response

response

{
  "data": [
    {
      "id": "nano-banana-2",
      "name": "Nano Banana 2",
      "type": "image",
      "capabilities": {
        "modes": ["text-to-image", "image-to-image"],
        "resolutions": ["1K", "2K", "4K"],
        "aspect_ratios": ["1:1", "16:9", "9:16", "4:3", "3:4"]
      },
      "pricing": {
        "default_credits": 12,
        "default_eur": 0.12,
        "strategy": "fixed"
      }
    },
    {
      "id": "kling-2.6-pro",
      "name": "Kling 2.6 Pro",
      "type": "video",
      "capabilities": {
        "modes": ["text-to-video", "image-to-video"],
        "duration_seconds": { "min": 5, "max": 10, "default": 5 },
        "aspect_ratios": ["16:9", "9:16", "1:1"]
      },
      "pricing": {
        "default_credits": 42,
        "default_eur": 0.42
      }
    }
  ]
}

Error Codes

Status	Label	Description
202	Accepted	Generation is still running. Poll the returned status_url.
400	Bad Request	Invalid request body or missing required fields.
401	Unauthorized	Missing or invalid API key.
402	Payment Required	Insufficient credits for the requested operation.
403	Forbidden	The API key does not have permission for this resource.
404	Not Found	The requested generation or asset could not be found for this API key.
429	Rate Limited	Too many requests. Back off and retry after the window resets.
500	Server Error	An unexpected error occurred on the server.
502	Provider Error	The upstream generation provider failed or returned an invalid response.
503	Temporarily Unavailable	A dependent service is temporarily unavailable; retry later.

Error Response Format

error response

{
  "error": "Human-readable error message",
  "code": "MACHINE_READABLE_CODE"
}

Rate Limiting

API requests are rate limited per API key. When the limit is exceeded, the server responds with HTTP 429. Inspect the following response headers to manage your request budget:

Header	Description
X-RateLimit-Limit	Max requests in the current window.
X-RateLimit-Remaining	Remaining requests in the current window.
X-Credits-Remaining	Credit balance after the request completes.