How to use the AI Storyboard Generator agent skill

What is an agent skill, and what does this one do?

An agent skill is a folder with a SKILL.md file that tells an AI agent when and how to perform a task. Once installed in the right directory, the agent reads the file and can use the skill when your request matches its description.

This skill covers storyboarding. Without any paid call, the agent can write a practical shot list from your script or brief. With your approval, it can also produce one to four key-frame images through Beatra. It does not produce finished video footage.

• Works from inside Claude Code, Codex or OpenClaw, per the skill's description.
• Shot lists need no paid call; key-frame images do.
• Not a video editor or video generator.

Understand what you are installing before you install it

The catalog entry you download from this page is a reviewed pointer, not the executable package. Its SKILL.md says it contains no client code and performs no Beatra operation by itself. The package it points to is a set of three standard-library Python scripts that make network calls and store a credential.

The pinned package has no self-updater, so it does not download code or change its own files. Its SKILL.md recommends reading the install and security sections before activating it. The license is listed as MIT in the catalog, while the SKILL.md front matter names MIT-0 (MIT No Attribution).

• Requires Python 3.10 or newer, standard library only, according to the package description.
• Requires Git to fetch the pinned package.
• Requires a Beatra account, browser sign-in to authorize, and paid Beatra credits for rendering.
• Runs scripts and uses the network.

Install the skill step by step

Step 1: Download the ZIP from this page and unzip it. Place the folder in ~/.claude/skills/ for all projects, or in a project's .claude/skills/ folder for one project. If you use the Claude apps instead, upload the skill under Settings, then Capabilities, then Skills.

Step 2: Fetch the pinned package into a review directory. The skill's install steps clone the repository and check out the pinned commit, then confirm the tree ID. Stop if the last check does not print TREE-OK. Approve this step only after you have read what it does.

Step 3: Inspect the files. The source lists checks for symlinks, executable bits and self-update code. These are inventory aids only; the SKILL.md says you must read every file in the pinned tree, including each script, before activation.

Step 4: Copy the reviewed folder into your skills directory. This changes agent configuration, so it needs a separate approval. The source refuses to overwrite an existing destination.

• Reading the grep output alone is not a source review.
• If you cannot review the Python scripts, stop and do not install.
• Start a new agent session if the host only discovers skills at startup.

Authorize Beatra and run a free connection check

Authorization opens a browser sign-in and stores a bearer token on your device. Only run it when you decide to connect. The source describes a free, non-billable check you can run first to confirm the connection works.

Ask for the connection check by name, or run it yourself from the installed folder.

• Authorization stores a bearer token.
• The verify command is non-billable according to the source.

How to trigger the skill with natural requests

You do not need special syntax. Describe the output you want, the source material, and that you accept a paid render if you want images. The agent should first show a shot list and a cost card, then wait for your explicit approval.

The source says that planning or a phrase like just do it does not count as approval. Be specific about what files, if any, you are willing to upload.

• Storyboard this 30-second ad brief into a shot list only, no images yet.
• Make four storyboard key frames for the attached scene script on Beatra. Show me the cost before you render.
• Check my Beatra connection and list the available storyboard models with their credit estimates.
• Plan a shot list for this short scene. Do not upload anything or make any paid call.

What happens during one paid render

The agent reads the installed SKILL.md and its references, then uses free discovery to read current model cards and credit estimates. It shows you a cost card with the tool, model, estimate and the files that will be uploaded.

After your approval, it uploads only the files you supplied for this task and submits once with a stable request ID. If a response is lost, it retries only with the same request ID and identical arguments. It then reports the artifact and the net charged credits, and asks again before any further paid stage.

• Check the cost card carefully before approving.
• Uploaded files and prompts go to a third party.
• Each further paid stage needs a new approval.

Security, safety and uninstall

The skill stores a bearer token, makes network calls and keeps local state under ~/.beatra. Do not use it for unreleased or confidential material you are not willing to send to a paid hosted service.

To remove it, run the uninstall script in dry-run mode first, read the JSON decision, then run it for real. The source says a disconnected result means the token was revoked when applicable and several Beatra state files were removed.

• Keep sensitive scripts and briefs out of uploads unless you have approved them.
• The package has no self-updater; to change versions, the source says to repeat the review steps for the new commit.
• An update command only reports what the installing store offers; for a manual copy it says to update the way you installed it.