media-generation

Q: How do I install media-generation?

Run npx skills add https://github.com/merit-systems/agentcash-skills --skill media-generation in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/media-generation/ (personal) or .claude/skills/media-generation/ (project).

by merit-systemsMedia & Creative

Generate images, videos, and SVGs using x402-protected AI models at StableStudio. USE FOR: - Generating images from text prompts - Generating videos from text or images - Editing images with AI - Vectorizing images to SVG - Creating visual content TRIGGERS: - "generate image", "create image", "make a picture" - "generate video", "create video", "make a video" - "edit image", "modify image" - "vectorize", "convert to svg" - "stablestudio", "nano-banana", "sora", "veo", "grok", "flux", "seedance", "gpt-image" ALWAYS use npx agentcash@latest fetch for stablestudio.dev endpoints.

Installs: 138
Last 24h: —
Trending rank: —
Security: Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/merit-systems/agentcash-skills --skill media-generation

01
Install
Run the command in your project. The CLI asks which agents to install for.
02
Describe the task
Ask your agent for something this skill covers. It loads media-generation automatically when relevant.
03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

Media Generation with StableStudio

Generate images and videos via x402 payments at https://stablestudio.dev.

Setup

If the agentcash CLI is not yet installed, see rules/getting-started.md for installation and wallet setup.

Pricing

All generation endpoints are dynamically priced async jobs. The static price advertises a range (up to $10; seedance up to $20) — the exact price depends on your options (quality, resolution, duration) and is quoted via x402 before payment, so the CLI shows the real cost per request. Verified sample quotes: gpt-image-2 $0.21 (defaults) / $0.01 (low quality, 1024x1024); flux-2-max $0.07 (defaults); seedance t2v $0.36 (4s, 480p).

Quick Reference

Image Models

Model	Endpoint	Cost	Time	Edit?
gpt-image-2 (default)	`gpt-image-2/generate`	$0.01-0.21 by quality/size	can take minutes	Yes
gpt-image-1.5	`gpt-image-1.5/generate`	dynamic	~3s	Yes
nano-banana-pro	`nano-banana-pro/generate`	dynamic	~10s	Yes
nano-banana	`nano-banana/generate`	dynamic	~5s	Yes
grok	`grok/generate`	dynamic	~3s	Yes
flux-2-pro	`flux-2-pro/generate`	dynamic	~5s	Yes
flux-2-max	`flux-2-max/generate`	~$0.07 at defaults	—	Yes

All endpoints are prefixed with https://stablestudio.dev/api/generate/. Edit endpoints use /edit instead of /generate.

Video Models

Model	Endpoint	Cost	Time
veo-3.1 (default)	`veo-3.1/generate`	dynamic	1-2min
veo-3.1-fast	`veo-3.1-fast/generate`	dynamic	~30s
grok-video	`grok-video/generate`	dynamic	~17s
seedance (Seedance 2 Pro)	`seedance/t2v` or `seedance/i2v`	dynamic, up to $20 ($0.36 for 4s/480p)	~1min
seedance-fast (Seedance 2 Fast)	`seedance-fast/t2v` or `seedance-fast/i2v`	dynamic, up to $20	~40s
wan-2.6	`wan-2.6/t2v` or `wan-2.6/i2v`	dynamic	2-5min
sora-2	`sora-2/generate`	dynamic	1-3min
sora-2-pro	`sora-2-pro/generate`	dynamic	2-5min

SVG Vectorization

Model	Endpoint	Cost
arrow-1.1	`arrow-1.1/vectorize`	dynamic
arrow-1.1-max	`arrow-1.1-max/vectorize`	dynamic

Image Generation

Default: gpt-image-2. This is the best default image model. Note: it can take minutes — poll the jobId and do not resubmit pending jobs. Use it unless the user specifically requests a different model or has a strong reason (e.g., speed, budget, or ultra-wide aspect ratios from grok).

npx agentcash@latest fetch https://stablestudio.dev/api/generate/gpt-image-2/generate -m POST -b '{
  "prompt": "a cat wearing a space helmet, photorealistic",
  "quality": "high",
  "size": "1536x1024"
}'

gpt-image-2 options:

quality: "low", "medium", "high" (affects cost: ~$0.01 low to ~$0.21 high)
size: "1024x1024", "1536x1024", "1024x1536", "auto"
background: "opaque", "auto". output_format: "png", "jpeg", "webp"

Other Image Models

gpt-image-1.5 -- Fastest GPT image model. Same schema as gpt-image-2; its background additionally supports "transparent":

npx agentcash@latest fetch https://stablestudio.dev/api/generate/gpt-image-1.5/generate -m POST -b '{
  "prompt": "a watercolor painting of a mountain lake",
  "quality": "high",
  "size": "1536x1024"
}'

Options: quality: "low"/"medium"/"high". size: "1024x1024", "1536x1024", "1024x1536", "auto".

nano-banana-pro -- High quality, supports up to 4K:

npx agentcash@latest fetch https://stablestudio.dev/api/generate/nano-banana-pro/generate -m POST -b '{
  "prompt": "a cat wearing a space helmet, photorealistic",
  "aspectRatio": "16:9",
  "imageSize": "2K"
}'

Options: aspectRatio: "1:1", "2:3", "3:2", "3:4", "4:3", "4:5", "5:4", "9:16", "16:9", "21:9". imageSize: "1K", "2K", "4K".

nano-banana -- Budget option. Adds extreme aspect ratios ("1:4", "4:1", "1:8", "8:1") on top of nano-banana-pro's, imageSize adds "512", and supports thinkingLevel: "minimal" (fastest) or "high" (best quality).

grok -- Fast, wide aspect ratio support. Note: uses aspect_ratio (underscore):

npx agentcash@latest fetch https://stablestudio.dev/api/generate/grok/generate -m POST -b '{
  "prompt": "neon cyberpunk cityscape at night",
  "aspect_ratio": "16:9"
}'

Options: aspect_ratio: "1:1", "16:9", "9:16", "4:3", "3:4", "3:2", "2:3", "2:1", "1:2", "19.5:9", "9:19.5", "20:9", "9:20".

flux-2-pro -- High quality with resolution control:

npx agentcash@latest fetch https://stablestudio.dev/api/generate/flux-2-pro/generate -m POST -b '{
  "prompt": "editorial photo of a vintage car in golden hour light",
  "aspect_ratio": "3:2",
  "resolution": "2 MP"
}'

Options: aspect_ratio: "1:1", "16:9", "9:16", "3:2", "2:3", "4:5", "5:4", "4:3", "3:4". resolution: "0.5 MP", "1 MP", "2 MP".

flux-2-max -- Same schema as flux-2-pro, higher quality tier; resolution additionally supports "4 MP".

SVG Vectorization

Convert any public image URL to SVG with arrow-1.1 (or arrow-1.1-max for higher quality):

npx agentcash@latest fetch https://stablestudio.dev/api/generate/arrow-1.1/vectorize -m POST -b '{
  "image": "https://example.com/logo.png"
}'

Optional: target_size (square resize target in pixels, 128-4096, default 1024), auto_crop (crop to dominant subject, default false). Poll the returned jobId for the SVG result.

Video Generation

Recommended: veo-3.1 (best quality)

npx agentcash@latest fetch https://stablestudio.dev/api/generate/veo-3.1/generate -m POST -b '{
  "prompt": "a timelapse of clouds moving over mountains",
  "durationSeconds": "6",
  "aspectRatio": "16:9"
}'

veo-3.1 options:

durationSeconds: "4", "6", "8"
aspectRatio: "16:9", "9:16"
resolution: "720p", "1080p"
negativePrompt: things to avoid

veo-3.1 also supports image inputs via imageMode ("none", "first-frame", "reference", "interpolation"): frame interpolation between image and lastFrame blob URLs, or up to 3 referenceImages for style guidance.

veo-3.1-fast -- Same schema as veo-3.1, cheaper, ~30s instead of 1-2min.

Job Polling

Generation returns a jobId and pollUrl. Poll until complete:

npx agentcash@latest fetch https://stablestudio.dev/api/jobs/{jobId}

Poll images every 3s, videos every 10s. Result contains imageUrl or videoUrl (SVG content for vectorize jobs). fetch handles both paid submission requests and free SIWX polling requests.

Other job endpoints (SIWX, free): GET /api/jobs lists all your jobs; DELETE /api/jobs/{jobId} soft-deletes a failed job.

Image Editing

Requires uploading the source image first. See rules/uploads.md.

All image models support /edit. The edit endpoint accepts images (array of blob URLs) plus a prompt.

npx agentcash@latest fetch https://stablestudio.dev/api/generate/gpt-image-2/edit -m POST -b '{
  "prompt": "change the background to a beach sunset",
  "images": ["https://...blob-url..."]
}'

Grok edit also supports aspect_ratio. GPT-image edit uses the same quality/size options as generate.

Model Comparison

Image Models

Model	Cost	Speed	Best For
gpt-image-2	$0.01-0.21	minutes	Best default; poll, don't resubmit
gpt-image-1.5	dynamic	~3s	Fast GPT images, transparent backgrounds
nano-banana-pro	dynamic	~10s	High quality, up to 4K resolution
nano-banana	dynamic	~5s	Quick drafts, extreme aspect ratios
grok	dynamic	~3s	Fast, many aspect ratios
flux-2-pro	dynamic	~5s	High quality, resolution control
flux-2-max	~$0.07+	—	Top Flux quality, up to 4 MP

Video Models

Model	Cost	Speed	Best For
veo-3.1	dynamic	1-2min	Best quality, interpolation, references
veo-3.1-fast	dynamic	~30s	Quick high-quality video
grok-video	dynamic	~17s	Cheapest short videos
seedance	up to $20	~1min	High quality, up to 15s, 4K upscale
seedance-fast	up to $20	~40s	Budget video with decent quality
wan-2.6	dynamic	2-5min	Budget, text or image input, audio sync
sora-2	dynamic	1-3min	Premium quality
sora-2-pro	dynamic	2-5min	Highest quality

FAQ

What does the media-generation skill do?

How do I install media-generation?

Run npx skills add https://github.com/merit-systems/agentcash-skills --skill media-generation in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/media-generation/ (personal) or .claude/skills/media-generation/ (project).

Which AI agents can use media-generation?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is media-generation safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains media-generation?

It is published by merit-systems in merit-systems/agentcash-skills. It has 138 recorded installs.