How to use the Multilingual Voiceover Dubbing agent skill

What the skill makes the agent do

An agent skill is a folder of instructions and scripts that your AI agent reads and follows when a task matches it. This one covers multilingual voice-over: you supply per-language scripts, and the agent helps cast voices for each market, runs pronunciation and timing pilots, and sorts the output by language and segment so a human can review each one.

The skill itself is a pointer to an external package. It does not perform any Beatra operation on its own. Once you install and authorize the package, the agent can use it to submit work to the paid Beatra service at mcp.beatra.ai.

• Plans per-language voice-over from your scripts
• Checks voices and timing with pilot samples before full runs
• Organizes delivered audio by language and segment
• Does not handle local or offline text-to-speech

What you need before you install

Check these requirements first. The install is multi-step and involves fetching code from GitHub and signing in through a browser.

• Git, to fetch the pinned package tree from github.com
• Python 3.10 or newer. The scripts use only the standard library and install no dependencies.
• A Beatra account with paid credits, and a browser for sign-in. The bearer token is stored locally.
• Network access to github.com and mcp.beatra.ai
• Comfort with a multi-step install. A one-click install is not available.

Install the skill from the ZIP or the Claude apps

Option one: download the ZIP from our page and unzip it so the folder is named ai-multilingual-dubbing and contains SKILL.md at its root. Put it in ~/.claude/skills/ for all projects, or in a project's .claude/skills/ folder for one project.

Option two: in the Claude apps, open Settings, then Capabilities, then Skills, and upload the package there.

Whichever route you use, the SKILL.md instructions ask for two separate approvals and a tree check before activation. Review every file in the package before you let the agent run anything. The package contains three Python scripts: scripts/authorize.py, scripts/mcp_client.py and scripts/uninstall.py.

• Approval one: fetch the pinned tree for review
• Approval two: copy the reviewed tree into your skills directory
• Authorization is a third, separate decision that opens a browser sign-in

Authorize the skill and run a free connection check

Authorization opens a browser sign-in and stores a bearer token on your machine. Only run it when you are ready to connect your Beatra account. Start a new agent session afterward if your host finds skills only at startup.

The connection check is free and non-billable. Use it to confirm the skill can reach the service before any paid work.

Ask for voice-over in plain language

Once installed and authorized, you do not need special syntax. Describe the job, the languages, the scripts and the review you want. The agent first uses free discovery to read current model cards and credit estimates, then shows you a cost card before any paid call.

A cost card lists the tool, the model, the estimated cost and the files that will be uploaded. Planning or saying just do it is not approval. Approve the card only for the exact payload shown.

• Example: Using the approved Spanish and German scripts, make pilot voice samples for each language and check timing before any full run.
• Example: Produce final voice-over for the three French segments, then sort the audio by language and segment for review.
• Example: List the available models and credit estimates for a 30-second English script. Do not submit anything yet.

Safety notes: scripts, network and stored data

This skill runs scripts, uses the network and stores a credential. Scripts and uploaded files are sent to a third party for hosted synthesis. Do not use it for sensitive or confidential scripts, and do not use it for local text-to-speech; the skill itself says not to.

The package has no self-updater and never downloads code by itself. The SKILL.md says a version update requires repeating the review steps for a new commit. The licence is listed as MIT in the catalog, while the SKILL.md front matter names MIT-0 (MIT No Attribution); check the LICENSE file in the pinned commit.

• Runs scripts: yes. Three standard-library Python scripts.
• Uses network: yes. Contacts github.com during install and mcp.beatra.ai for work.
• Stores a bearer token locally after sign-in.
• Sends scripts and uploaded files to a paid third-party service.
• Requires a cost estimate and explicit approval before each paid call.

Tips for reliable results and clean removal

Use pilot samples before full runs so you catch wrong voices or timing problems cheaply. Review each language separately, since each market has its own voice and pronunciation checks.

To uninstall, run a dry run first and read the JSON decision. The uninstall script reports whether this was the last Beatra skill on the device; if so, it revokes the token and removes the local Beatra credential and state files.

• Keep one approved payload per paid call, and check the billing figure after each stage.
• If a response is lost, retry only with the same client_request_id and identical arguments.
• Never start a replacement submission while a task is still running.