ClaudeCursorCodexContinueCline

Stensyl, inside any AI agent

Connect Stensyl to your workflow and generate images, video, audio, and 3D straight from your prompts. One connection, every model, on your own account and your own credits.

See pricing
Terminal
> stensyl image "dusk hero, faceted facade"
--model gpt-image-2 --out hero.png
Ray
Rendering…
saved ~/hero.png
Three front doors

Connect in under a minute

Pick MCP, the CLI, or a skill. Every door signs in with your account and runs on your credits.

Connect Stensyl as a tool inside any MCP-compatible agent.

1

Open your agent's connector settings

Claude (web, mobile, or Desktop), Cursor, Continue, or any MCP-compatible agent.

2

Add the Stensyl connector

Paste the server URL into your agent's custom connector field.

https://mcp.stensyl.ai/api/mcp
3

Sign in with your Stensyl account

Approve the consent screen. Generations run on your own credits, then just ask your agent to make something.

Using Claude Code, Codex, or another terminal agent? The CLI is the smoothest path.
In practice

Just describe it, Stensyl makes it

Real commands, real output. A look at what your agent does when you ask.

Terminal

Generate from your terminal

Batch a product set or a whole hero grid in one command. Loop over a list, pipe the output into your build, all on your own credits.

Terminal
> for shot in hero closeup wide detail; do
stensyl image "$shot, studio product light" \
--model gpt-image-2 --out shots/$shot.png
> done
Ray
Queued 4 jobs · gpt-image-2
hero.png 2048×2048
closeup.png 2048×2048
wide.png 2048×2048
detail.png 2048×2048
4 images · saved to ./shots
Any agent

Drop Stensyl into your agent

One connection exposes every model. Ask in plain language and the agent picks the right tool, or name the model yourself. No keys to juggle.

Terminal
> Generate a 5s dusk flythrough of the facade,
then a square teaser image for social
Ray
Two jobs from one ask:
generate_video kling-3-pro · 5s · 16:9
generate_image gpt-image-2 · 1:1
→ flythrough.mp4
→ teaser.png
Image & video

Stills to cinematic video

From a quick draft image to a finished clip. Your model, your parameters, up to 4K, any aspect, any duration the model supports.

Terminal
> stensyl video "slow push-in on the atrium at golden hour" \
--model kling-3-pro --duration 5
Ray
Submitted job · kling-3-pro · 5s
Polling 7f3a92… rendering
clip ready 1080p · 5s
saved ~/atrium.mp4
Long renders run async · poll, fetch, done
The surface

The command set

The full tool surface your agent can call once connected.

account

Your plan, credit balance, and recent usage — read from the agent.

plancreditsusage

models

The live catalogue: every model, its parameters, limits, and cost.

listinspect

generate

The workhorse. Submit a job, track progress, fetch the result.

imagevideoaudio3d

write

Draft and rewrite with Stensyl's writing models.

draftrewrite

transcribe

Speech to text from any audio or video file.

audiovideo

jobs

Poll a long render and pull the output when it's ready.

statusfetch
Questions

Good to know

Connect plugs Stensyl into your AI agent or terminal three ways: as an MCP connector, a CLI, or a skill. You sign in with your own Stensyl account, and everything you generate runs on your normal credits. There are no separate developer keys to manage.

Any MCP-compatible agent, including Claude (web, mobile, and Desktop), Cursor, Continue, and Cline. The CLI and skills work with Claude Code, Codex, and any agent that supports skills, plus any terminal, script, or CI pipeline.

Every model on Stensyl: image, video, audio, and 3D, across all our providers. Ask in plain language and the agent picks a suitable model, or name the exact one you want.

Images, video, audio, 3D, and text. Stills up to 4K, video at the duration and aspect the model supports. Same models and quality as the web app.

No. You sign in with your Stensyl account: a quick browser login for the CLI, or an OAuth consent screen when you connect an agent. Nothing to copy, paste, or rotate.

Every call bills your normal Stensyl credit balance at the same rate as the web app. No separate metering, no surcharge. A free account can set everything up and explore the commands; generating needs Starter or higher.

It depends on the model and what you are making, exactly as it does in the web app. Longer renders run in the background, so your agent can carry on and fetch the result when it is ready.

Yes, for models that support it. Reference images, starting frames for image-to-video, and edits all take an input you provide.

Yes. Any agent that speaks MCP connects with the server URL, and the CLI runs anywhere you have a terminal. If you build your own tooling, the CLI is the smoothest path.

Your agent just got a studio

Connect Stensyl to Claude, Cursor, Codex, or any MCP agent in under a minute. Image, video, audio, and 3D on demand, on your own credits.