Is there a Claude Code skill for AI video generation?
Yes, Wireflow's AI video generation skill for Claude Code plans the shot, quotes about 40 credits, then renders a 5 second 16:9 clip from text or an image.
Yes, Wireflow's ai-video-generation skill turns a text prompt or a start image into a 5 second, 16:9 video clip inside Claude Code. It plans the shot, shows the credit cost and waits for your yes, then renders the clip with MiniMax Hailuo 03 Max Turbo and saves the MP4 to ./wireflow-outputs/.
The planning happens in Claude, and the render runs as a published Wireflow app over the Wireflow MCP connector, signed in as you. Without the connector, the skill still writes a shot list and prompts that work in any video model. Skill files and examples are in the skill's folder on GitHub.
How do I install the video skill in Claude Code?
Install the Wireflow plugin, which adds every Wireflow skill and the Wireflow connector:
/plugin marketplace add wireflowINC/skills
/plugin install wireflow-skills@wireflow
Claude Code shows the plugin's details first, so you can review it and choose its scope. Then run /mcp, choose wireflow and sign in with your Wireflow account. The skill shows up as /wireflow-skills:ai-video-generation, and Claude also picks it up from plain requests.
To add only this skill, use the skills CLI, or copy the skills/ai-video-generation folder into ~/.claude/skills/. Then connect Wireflow once and run /mcp to sign in. The connector uses OAuth, so there is no API key to copy.
npx skills add wireflowINC/skills --skill ai-video-generation
claude mcp add --transport http wireflow https://www.wireflow.ai/api/mcp
How do I use the skill to generate a video?
Describe the clip in plain words, and Claude plans it before anything renders:
Make a 5 second b-roll clip for a coffee shop reel: steam rising from a cup
on a rainy windowsill, slow push-in, cozy and moody.
Requests like "text to video", "animate this image", "make a short clip of..." or "I need b-roll for..." also start the skill. From there it:
- Writes each shot as one subject, one action, one camera move, the light and the style, sized to fit five seconds.
- Shows the shot list, the credit cost and your balance, and waits for your yes.
- Renders each shot with
run_appand pollsget_executionuntil it finishes. - Saves each MP4 to
./wireflow-outputs/, and pulls a few stills to check the clip if ffmpeg is installed. - Reports the file path, the link and the credits used.
Every extra take needs its own yes.
What kind of clip does it make, and does it have sound?
Each shot becomes one MP4: 5 seconds, 16:9, about 1344x768, with an audio track the model adds.
In the skill's recorded test on 8 October 2026, this text-to-video prompt made a 1344x768 MP4 of 5.2 seconds with an audio track, for 40 credits:
A small paper boat drifts across a rain puddle on a city street at night. Slow dolly-in at ground level. Neon shop signs in pink and teal reflect and ripple in the water, light rain falls, shallow depth of field, cinematic, moody
You can watch that clip from the skill's examples page.
Can it turn my own image into a video?
Yes, give it a start image at a public https:// URL and the clip opens on that frame. With a start image, the shot only needs to say what moves and how the camera moves, because the image already sets the look.
For a file on your computer, the Wireflow connector can send you an upload link, and the skill uses the uploaded file's URL. An image you made with Wireflow works directly. To give several clips one look, make a still with the ai-image-generation skill and use it as the start image for each clip.
Can it make vertical or longer videos?
Not as installed: the skill's app makes 16:9 clips of 5 seconds. For a vertical 9:16 or a longer clip, open the skill's board in Wireflow from the link the skill gives you, duplicate it, and change the ratio, length or model in your copy.
Text and logos inside a generated video come out garbled, so add them later in your editor.
Can it edit a video I already have?
No, the skill only makes new clips, and it is not for cutting, captioning or editing footage that already exists. For that, see how to connect Claude to video editing.
How many credits does one AI video cost?
About 40 credits per clip, which is 8 credits a second, as listed in the skills repo on 8 October 2026.
| What | Credits |
|---|---|
| Shot planning, prompts and the quote | 0 |
| One rendered clip (5 s, 16:9, 768P) | About 40 |
| A second take | About 40 more, only with your yes |
Rendering needs credits on your Wireflow account. The Free plan includes 0 credits each month, but new accounts get a free video; more videos need a paid plan, and plans start at $36 a month for 2,660 credits (pricing). The skill's README puts that at roughly 40 clips.
Does the video skill work in Cursor, Codex or Claude Desktop?
It is built for Claude Code, and the skills CLI also installs it in Cursor, Codex and other agents that read SKILL.md. Rendering then needs that agent connected to the Wireflow MCP server, and the skill's README says Wireflow has not tested those agents itself. In Claude.ai and Claude Desktop, add https://www.wireflow.ai/api/mcp as a custom connector; saving files needs a shell, so there you get the video link instead.
Connection steps for each agent are in Use Wireflow from your agent. To reach Wireflow's other video models from an agent without this skill, see AI video generation MCP. Every Wireflow skill is listed on the agent skills page.