Revana skills and examples

Agent skills, runnable examples and n8n workflows for Revana: one API key for 200+ image, video, music and chat models, including Seedance 2.5, Veo 3.1, Kling, Wan, Hailuo, MiniMax, Nano Banana Pro, GPT Image, Flux, Midjourney, Suno, and 90+ LLMs through an OpenAI-compatible endpoint.

• One request shape for every model: POST /v1/generations with model and input, poll, download.
• Cheapest route first, automatic fallback. You pay only for the route that succeeds. Failed generations cost nothing.
• Hard price cap per job (max_cost_usd) and idempotent retries (Idempotency-Key).
• Prepaid credits: no subscription, no monthly quota. Prices: https://revana.si/pricing

Install the skills

Works with Claude Code, Codex, Cursor and other agents that read the open Agent Skills format.

Tool	Install
Any agent	npx skills add Aribk7/revana-skills
Claude Code	npx skills add Aribk7/revana-skills -a claude-code, or /plugin marketplace add Aribk7/revana-skills then /plugin install revana-skills@revana-skills
Codex	npx skills add Aribk7/revana-skills -a codex
Cursor	npx skills add Aribk7/revana-skills -a cursor
By hand	copy a folder from skills/ into your agent's skills directory (e.g. ~/.claude/skills/)
MCP instead	claude mcp add --transport http revana https://revana.si/api/mcp (other apps)

Then set your key in the shell your agent runs in (create one at https://revana.si/dashboard):

export REVANA_API_KEY=sr_live_...

and ask: "make a 5 second video of a paper boat in the rain with seedance", "what's the cheapest veo api?", "make a UGC ad for this product photo".

Skills

Skill	Use it for	Ask it like
revana-video	Text or image to video: Seedance 2.5, Veo 3.1, Kling, Wan, Hailuo, MiniMax, Grok Imagine, Vidu	"make a video with seedance", "veo 3 api", "animate this image"
revana-image	Make and edit pictures: Nano Banana Pro, GPT Image 2, Flux, Seedream, Imagen, Midjourney	"nano banana pro api", "edit this photo", "product shot"
revana-ugc-ad	A short product ad: a still from your product photo, then a 9:16 talking video	"make a UGC ad", "TikTok ad for my product"
revana-model-picker	The cheapest model for a job, ranked from live prices (bundled script, no key)	"cheapest veo api", "cheapest text to video"
revana-music	Songs and instrumentals with Suno V6	"make a song", "suno api"

Every skill quotes the price (from the server where it can, else the live list) and waits for your OK before spending, sends max_cost_usd at the agreed price, makes one Idempotency-Key per generation, never prints your key, and works with the revana CLI or plain curl.

Examples

examples/: runnable scripts that call https://api.revana.si.

curl	text to video, image to video with an upload, chat, free calls
Node	text to video, image to video, batch images, chat (plain fetch), and the revana SDK
Python	the same, standard library only, and the revana package
Next.js	route handlers: server-side key, model allowlist, spend cap, polling, file redirect
Vercel AI SDK	tools: list models, quote, generate (with approval), status
MCP	remote MCP config for Claude Code, Codex, Cursor, VS Code, ChatGPT

n8n

n8n/: importable workflows, AI video generator with Seedance 2.5 via Revana and Product photo to UGC video ad. Each one quotes from live prices, creates the job with a unique Idempotency-Key, polls with a Wait node, and downloads the video. How to import.

The API in 30 seconds

# Models, options and prices (no key)
curl https://api.revana.si/v1/models

# Make a video. One Idempotency-Key per video, made once: resend with the same $KEY after a timeout or a 5xx
KEY=$(uuidgen)
curl https://api.revana.si/v1/generations \
  -H "Authorization: Bearer $REVANA_API_KEY" -H "Content-Type: application/json" -H "Idempotency-Key: $KEY" \
  -d '{"model": "seedance-2.5", "input": {"prompt": "a paper boat drifting down a rainy street", "resolution": "720p", "duration": 5}, "max_cost_usd": 1.05}'

# Poll, then download
curl https://api.revana.si/v1/generations/$ID -H "Authorization: Bearer $REVANA_API_KEY"
curl -L https://api.revana.si/v1/generations/$ID/content -H "Authorization: Bearer $REVANA_API_KEY" -o out.mp4

Prices change, so read them from /v1/models when you need them. As a dated example, on 2026-10-31 that 5-second 720p Seedance 2.5 clip listed at $1.05 ($0.21 a second). With a start image (image_url) Seedance 2.5 costs more than the listed price; on API versions deployed after 2026-10-31, POST /v1/generations/quote prices the exact request for free (an older server answers 404).

Retrying. After a timeout, a dropped connection, a 5xx or a 429 on the create, resend the same body with the same Idempotency-Key: you get the generation already started, never a second charge. After a generation has failed, start again with a new key: the old one keeps answering the failed generation. When asking for help, quote the request_id from the error (or the X-Request-Id header, on API versions deployed after 2026-10-31).

Also on the same key:

• POST /v1/uploads: a public link for a local file, to use as a reference (free)
• GET /v1/balance: what the key can spend
• POST /v1/chat/completions: OpenAI-compatible chat. Point any OpenAI SDK at https://api.revana.si/v1.

SDKs: npm i revana (npx revana generate seedance-2.5 "..." -o out.mp4), and pip install revana.

Docs: https://revana.si/docs · OpenAPI: https://revana.si/openapi.json · For LLMs: https://revana.si/llms.txt · Full reference: https://revana.si/llms-full.txt

License

MIT