
Agent skill for Claude Code, Cursor and Codex to generate adult images, NSFW image-to-video and
Agent skill for Claude Code, Cursor and Codex to generate adult images, NSFW image-to-video and uncensored edits via SpicyAPI (18+).
·
English · 日本語 · 한국어 · Français · Español · Русский
An agent skill that lets Claude Code, Cursor, Codex, Windsurf, Gemini CLI and OpenClaw generate adult (18+) images, NSFW image-to-video clips, uncensored image edits and uncensored text, through one pay-per-use API.
Install · What you can ask · Models & prices · How it works · CLI · FAQ
Real outputs from the models this skill calls (Wan 2.2 Spicy, Seedance 2.0 Spicy, Wan 2.7 Spicy). Click for the full clip.
18+ only. The skill refuses sexual content involving minors or anyone who appears to be a minor, sexual content of real people without documented consent (including face swaps and "undressing" photos), and impersonation. It writes every prompt with an explicit adult age.
spicy.py models.Sign up at spicyapi.ai, top up (card, Apple Pay / Google Pay, or crypto), then create a key in the Console. Set a spend cap on the key if you like.
export SPICY_API_KEY="sk-spicy-..."Any agent (skills CLI: Claude Code, Cursor, Codex, Windsurf, Gemini CLI, OpenClaw and more)
npx skills add Spicy-API/nsfw-ai-skillClaude Code plugin marketplace
/plugin marketplace add Spicy-API/nsfw-ai-skill
/plugin install nsfw-ai@spicyapi-nsfwManual
git clone https://github.com/Spicy-API/nsfw-ai-skill.git
cp -r nsfw-ai-skill/skills/nsfw-ai ~/.claude/skills/ # Claude Code
# or: cp -r nsfw-ai-skill/skills/nsfw-ai ~/.codex/skills/ # Codex
# or: point your agent's skills directory at skills/nsfw-aiThe skill works on its own. If your client supports MCP, you can also add SpicyAPI's MCP server for model listing, quotes and task management:
claude mcp add spicyapi -e SPICY_API_KEY=$SPICY_API_KEY -- npx --yes --package=@spicyapi/mcp spicyapi-mcpCursor / Claude Desktop / Windsurf (mcp.json):
{
"mcpServers": {
"spicyapi": {
"command": "npx",
"args": ["--yes", "--package=@spicyapi/mcp", "spicyapi-mcp"],
"env": { "SPICY_API_KEY": "sk-spicy-..." }
}
}
}| You say | The skill does |
|---|---|
| "List the Spicy video models and what they cost." | Reads the live catalog (no key needed) |
"Animate ./frame.jpg into a 5-second clip: she turns toward the camera, candlelight, slow push-in. Cheapest option." | Uploads the image, uses Wan 2.2 Spicy at 480p, quotes ~$0.095, waits for your OK, saves the MP4 |
| "Photoreal first frame: a woman in her 30s in red lingerie on satin sheets, then animate it." | Qwen Image 2.1 → Wan 2.2 Spicy or Seedance, two quotes |
"Re-render that at 720p on Seedance 2.0 Spicy with a last frame from end.jpg." | Seedance 2.0 Spicy with last_image_url |
| "Anime still of an adult demon queen, then make it move." | Prefect Pony XL → Vidu Q3 Spicy |
"Change the outfit in me.png to a black satin slip dress." | Qwen Image 2.1 Edit (you, a consenting adult, or a fictional character only) |
| "Write three NSFW video prompts for this frame and estimate each." | Prompt recipe + quotes, nothing runs until you choose |
| "Extend my last clip by 5 seconds with the same LoRA." | Wan 2.2 Spicy LoRA video-extend |
| "Four seed variations at 480p, then the best one at 720p." | Batch with a total price confirmation |
More ideas in examples/prompts.md. 100+ ready prompts in nsfw-ai-video-prompts.
The skill can call any model in the SpicyAPI catalog. These are the ones that matter for adult work, in catalog order (most popular first, newest version first), with the Spicy Index and Freedom Score from the public SpicyAPI leaderboards (✅ Freedom 90+ · ◐ 70–89 · ⚠️ under 70 · 🧪 under 15 test runs). 🌶️ Spicy editions are tuned for adult output; Standard models here have the catalog tier unrestricted, so they take mature prompts too and add text-to-video and reference-to-video. Cheapest tier shown; the skill always shows the exact quote before running. Catalog read on
2026-09-27
Video
| Model | Type | Tasks | Duration | From | Spicy Index | Freedom |
|---|---|---|---|---|---|---|
| Seedance 2.5 Spicy | 🌶️ Spicy | I2V | 4–30 s | $0.216/s | 56.5 | ✅ 96.7 |
| Seedance 2.5 | Standard | I2V, Ref2V, T2V | 4–30 s | $0.1234/s | 69.5 | ◐ 80.9 |
| Seedance 2.0 Spicy | 🌶️ Spicy | I2V | 4–15 s | $0.114/s | 61.5 | ✅ 93.3 |
| Seedance 2.0 | Standard | I2V, Ref2V, T2V | 4–15 s | $0.07/s | 81.5 | ◐ 70.4 |
| Wan 3.0 Prime | Standard | I2V, Ref2V, T2V | 2–30 s | $0.0612/s | 76.5 | ◐ 78 |
| Wan 3.0 | Standard | I2V, Ref2V, T2V | 2–30 s | $0.045/s | 76.5 | ✅ 96 |
| MiniMax H3 Spicy | 🌶️ Spicy | I2V | 3–15 s | $0.038/s | 29.5 | ✅ 97.5 |
| MiniMax H3 | Standard | I2V, Ref2V, T2V | 4–15 s | $0.025/s | 72.5 | 🧪 33.3 |
| MiniMax H3 Singularity LoRA | Standard | I2V, Ref2V | 3–15 s | $0.06/s | 72.8 | ✅ 100 |
| LTX 2.5 | Standard | I2V, T2V | 5–20 s | $0.09/s | 66 | ◐ 80.3 |
| Wan 3.0 Pro Prime | Standard | I2V, Ref2V, T2V | 2–30 s | $0.234/s | 76.5 | ◐ 82 |
| Wan 3.0 Pro | Standard | I2V, Ref2V, T2V | 2–30 s | $0.144/s | 76.5 | ◐ 82 |
| MiniMax H3 LoRA | Standard | I2V, Ref2V, T2V | 3–15 s | $0.05/s | 75.2 | ✅ 98.3 |
| HappyHorse 1.1 | Standard | I2V, Ref2V, T2V | 3–15 s | $0.14/s | 62.5 | ⚠️ 65.1 |
| Seedance 2.0 Mini Spicy | 🌶️ Spicy | I2V | 4–15 s | $0.0387/s | 44.5 | ✅ 93.3 |
| Seedance 2.0 Mini | Standard | I2V, Ref2V, T2V | 4–15 s | $0.01097/s | 64.5 | ⚠️ 64.9 |
| Wan 2.7 Spicy | 🌶️ Spicy | I2V | 2–15 s | $0.1235/s | 46.5 | ✅ 100 |
| LTX 2.3 Spicy | 🌶️ Spicy | I2V | 3–20 s | $0.019/s | 33.5 | ◐ 89.2 |
| LTX 2.3 Spicy LoRA | 🌶️ Spicy | I2V | 3–20 s | $0.0285/s | 34.8 | ◐ 83.8 |
| Seedance 2.0 Fast Spicy | 🌶️ Spicy | I2V | 4–15 s | $0.081/s | 44.5 | ✅ 90 |
| Seedance 2.0 Fast | Standard | I2V, Ref2V, T2V | 4–15 s | $0.02254/s | 64.5 | ⚠️ 68.2 |
| Vidu Q3 Turbo | Standard | I2V | 1–16 s | $0.042/s | 39.5 | ✅ 93.3 |
| Vidu Q3 Spicy | 🌶️ Spicy | I2V | 1–16 s | $0.0665/s | 46.5 | ✅ 96.7 |
| Vidu Q3 | Standard | I2V | 1–16 s | $0.07/s | 46.5 | ✅ 93.3 |
| Vidu Q3 Pro | Standard | I2V | 1–16 s | $0.054/s | 36.5 | ✅ 93.3 |
| Seedance 1.5 Pro Spicy | 🌶️ Spicy | I2V | 4–12 s | $0.012/s | 48.5 | ✅ 96.7 |
| Seedance 1.5 Pro | Standard | I2V, T2V | 4–12 s | $0.0112/s | 46 | ✅ 90 |
| Wan 2.6 Flash | Standard | I2V | 5, 10, 15 s | $0.0225/s | 31.5 | ✅ 100 |
| Wan 2.6 Spicy | 🌶️ Spicy | I2V | 5, 10, 15 s | $0.095/s | 46.5 | ✅ 96.7 |
| Wan 2.6 | Standard | I2V, Ref2V, T2V | 5, 10, 15 s | $0.065/s | 58.5 | 🧪 8.7 |
| Wan 2.5 | Standard | I2V, T2V | 5, 10 s | $0.045/s | 46 | ✅ 99 |
| Wan 2.2 Spicy | 🌶️ Spicy | I2V | 5, 8 s | $0.019/s | 23.5 | ✅ 91.2 |
| Wan 2.2 Spicy LoRA | 🌶️ Spicy | I2V, Extend | 5, 8 s | $0.024/s | 25 | ◐ 74.8 |
| Wan 2.2 LoRA | Standard | I2V | 5, 8 s | $0.024/s | 22.5 | ◐ 88.8 |
Images (recommended: Qwen Image 2.1, from $0.024)
| Model | Type | Tasks | From | Spicy Index | Freedom |
|---|---|---|---|---|---|
| Qwen Image 2.1 | Standard | Edit, T2I | $0.024/image | 73 | ✅ 96.3 |
| Qwen Image 2.1 LoRA | Standard | Edit, T2I | $0.03/image | 80.5 | ✅ 92 |
| MiniMax H3 Image LoRA | Standard | Edit, T2I | $0.042/image | 74.5 | ✅ 100 |
| Qwen Image 3.0 Pro | Standard | Edit, T2I | $0.04/image | 56 | ✅ 98 |
| Qwen Image 3.0 | Standard | Edit, T2I | $0.03/image | 56 | ✅ 96 |
| Seedream 5.0 Pro | Standard | Edit, T2I | $0.036/image | 73 | ✅ 94.3 |
| Qwen Image Edit Spicy | 🌶️ Spicy | Edit | $0.038/image | 14 | ✅ 96 |
| Seedream 5.0 Lite | Standard | Edit, T2I | $0.0345/image | 73 | ✅ 96 |
| Qwen Image 2 | Standard | Edit, T2I | $0.035/image | 34 | ✅ 96.7 |
| Qwen Image 2512 LoRA | Standard | Edit, T2I | $0.03/image | 50.5 | ✅ 92.5 |
| Z-Image Spicy Pro | 🌶️ Spicy | T2I | $0.019/image | 38 | ✅ 100 |
| Z-Image Spicy | 🌶️ Spicy | T2I | $0.01235/image | 32 | ✅ 98.8 |
| Z-Image | Standard | T2I | $0.01/image | 17 | ✅ 100 |
| Z-Image Turbo LoRA | Standard | Edit, T2I | $0.012/image | 46.5 | ✅ 95 |
| Seedream 4.0 | Standard | Edit, T2I | $0.03/image | 74 | ◐ 74.7 |
| Prefect Pony XL | Standard | T2I | $0.015/image | 30 | 🧪 36 |
| FLUX.1 Dev LoRA | Standard | T2I | $0.018/image | 32.5 | ◐ 75 |
Text (OpenAI-compatible chat command): Grok 4.7, DeepSeek V4.1 Flash and other models with catalog tier unrestricted, from $0.0012 per 1K tokens.
Full field-level reference: skills/nsfw-ai/references/models.md.
You ──► Agent reads SKILL.md
│ 1. checks the request against the adults-only / consent rules
│ 2. picks a model (or lists live ones) spicy.py models --spicy
│ 3. reads the live input schema spicy.py schema <model>
│ 4. writes the prompt (references/prompting.md)
│ 5. quotes and asks you to approve spicy.py generate ... → needs_confirmation
│ 6. runs after your OK spicy.py generate ... --yes
▼
SpicyAPI /api/v1/jobs/quote → /jobs/createTask (Idempotency-Key) → /jobs/recordInfo
▼
./spicy-output/<taskId>_0.mp4spicy:// URIs; public HTTPS URLs are passed straight through.quoteId + expectedCost), so the settled charge can never exceed the price you approved.The same script is a normal CLI:
S=skills/nsfw-ai/scripts/spicy.py
python3 $S models --spicy --modality video # no key needed
python3 $S schema alibaba/wan-2.2-spicy/image-to-video
python3 $S generate alibaba/qwen-image-2.1/text-to-image \
-p "Photorealistic boudoir portrait of a woman in her early 30s in black lace lingerie, window light" \
--set aspect_ratio=2:3 --set resolution=1k --max-cost 0.03
python3 $S generate alibaba/wan-2.2-spicy/image-to-video \
--image ./spicy-output/<taskId>_0.png \
-p "She turns slowly toward the camera, lace strap slipping, warm lamp light, slow push-in" \
--set duration_seconds=5 --set resolution=480p --yes
python3 $S status <taskId> --wait --download ./spicy-output
python3 $S chat xai/grok-4.7/chat "Write a 60-word image-to-video prompt for a rainy-window boudoir scene, woman in her 30s"Run the tests with python3 -m unittest discover tests. Example pipelines: examples/image-to-video-pipeline.sh, examples/batch-variations.sh.
Prefer an SDK? SpicyAPI ships official ones: npm install @spicyapi/sdk, pip install spicyapi, go get github.com/Spicy-API/spicy-go, composer require spicyapi/spicyapi, plus @spicyapi/cli. See the developer docs.
nsfw-ai-skill/
├── skills/nsfw-ai/
│ ├── SKILL.md # instructions the agent loads
│ ├── scripts/spicy.py # zero-dependency SpicyAPI CLI
│ ├── references/models.md # model IDs, fields, price tiers
│ ├── references/prompting.md # prompt recipe, negatives, couples, anime
│ └── agents/openai.yaml # Codex / OpenAI agents metadata
├── .claude-plugin/ # Claude Code plugin + marketplace manifests
├── examples/ # pipelines and things to ask
└── tests/ # offline unit testsClaude Code itself does not generate media. With this skill installed, Claude Code (and Cursor, Codex, Windsurf, Gemini CLI, OpenClaw) calls SpicyAPI's Spicy models, which do allow adult output, and saves the files locally. The skill keeps hard limits: adults only, no real people without consent.
On SpicyAPI's catalog (2026-09-27) the cheapest models that pass the explicit tests are Seedance 1.5 Pro Spicy (from $0.012/s at 480p, Freedom 96.7) and Wan 2.6 Flash ($0.0225/s at 720p, Freedom 100). For the best result per dollar, Wan 3.0 costs $0.45 per 5-second 720p clip with Freedom 96.
The skill is free and open source (MIT). Generation is paid per output on SpicyAPI because GPUs cost money; there is no subscription and failed tasks are refunded. For free generation, run open-weight models locally (see awesome-nsfw-ai).
Yes. The Spicy editions are the default for adult requests, but spicy.py accepts any model ID in the SpicyAPI catalog: standard Seedance / Wan / MiniMax video models, Seedream / Qwen / Wan image models, upscalers, face and lip-sync tools, and text models via chat. Run spicy.py models (without --spicy) to list them.
Yes. Use the skill alone, or add the official SpicyAPI MCP server (@spicyapi/mcp) alongside it.
Outputs download to ./spicy-output/. Output links expire after about 20 minutes. SpicyAPI keeps prompts, uploads and outputs on separate retention clocks that you can shorten, and you can destroy a finished task's content; see Trust.
Qwen Image 2.1 LoRA with an anime LoRA for the still (adult character design; #1 on the image leaderboard, Freedom 92), then Vidu Q3 Spicy (Freedom 96.7) or Wan 2.2 Spicy LoRA with the same LoRA for motion. Prefect Pony XL works with tag-style prompts but has little test data so far.
The model provider may still refuse some inputs; that shows up as a failed task and the charge is refunded. Change the wording or try another Spicy model. A 40004 error means the chosen parameter combination is not offered; change the named field.
Adults only. No sexual content involving anyone under 18 or who appears under 18, in any style. No sexual content of real people without documented consent, no face swaps into sexual content, no "undressing" real photos. Follow the law where you and your audience are. SpicyAPI's Content Policy and Acceptable Use Policy apply to every request.
⭐ Star the repo if it saved you an afternoon of API docs.
Pick your client and paste the snippet. Each one is the same server, written the way that client expects it.
claude mcp add spicyapi -- npx --yes --package=@spicyapi/mcp spicyapi-mcp{
"mcpServers": {
"spicyapi": {
"env": {
"SPICY_API_KEY": ""
},
"args": [
"--yes",
"--package=@spicyapi/mcp",
"spicyapi-mcp"
],
"command": "npx"
}
}
}code --add-mcp '{"name":"spicyapi","env":{"SPICY_API_KEY":""},"args":["--yes","--package=@spicyapi/mcp","spicyapi-mcp"],"command":"npx"}'[mcp_servers.spicyapi]
command = "npx"
args = ["--yes", "--package=@spicyapi/mcp", "spicyapi-mcp"]Runs locally on your device. Your client starts the server itself, so nothing has to be hosted.
Paste this prompt into your agent. It reads this page and does the setup for you.
Read https://aiagentslisting.com/mcp/nsfw-ai-skill to learn what the "Nsfw Ai Skill" MCP server does and how to install it. Add it to my coding agent's MCP configuration as documented on that page, then confirm the server connects and list the tools it exposes.Agents can also browse this directory over MCP at https://aiagentslisting.com/api/mcp
npx skills add Spicy-API/nsfw-ai-skillclaude mcp add spicyapi -e SPICY_API_KEY=$SPICY_API_KEY -- npx --yes --package=@spicyapi/mcp spicyapi-mcpThis server runs locally, so we can't read its tool list over the web yet.
We could not run this server: its npm package has not answered in our offline sandbox, which usually means it needs an API key or other configuration to start. The four checks that need a live server are left out, and the grade tops out at B until we can read its tools.
Maintain this server? Claiming proves you own the listing and earns the last 4 points.
Claim this listingPaste this line near the top of your repository's README. The badge always renders the current score.
[](https://aiagentslisting.com/mcp/nsfw-ai-skill)Nothing comparable is listed yet.
No tagged releases on record.
Connect to the AI Agents Listing MCP server at https://aiagentslisting.com/api/mcp using the HTTP transport and complete the sign-in when it asks. Then call the submit_review tool for the listing "nsfw-ai-skill" with a rating from 1 to 5 and a short note on what worked and what did not.No reviews yet
Be the first to share your experience with this tool.
One email a week. New agents, MCP servers and skills, and what is actually getting traction.