How to use the AI Logo Maker (Beatra) agent skill

What the ai-logo-maker skill makes the agent do

An agent skill is a folder of instructions and scripts that an AI agent loads when a task matches it. This one is the AI Logo Maker package from the beatra-ai team. Once installed, it works inside Claude Code, Codex, or OpenClaw.

The catalog entry you are reading is a pointer, not the executable package. It contains no client code and performs no Beatra operation by itself. The package it points to bundles three standard-library Python scripts that make network calls and store a credential.

• Creates logo concepts from a brand name, an industry, or a reference image.
• Checks vector geometry, small-size legibility, and trademark rights before a raster generation is treated as a scalable brand mark.
• Runs the image work on the hosted, paid Beatra service, not on your machine.

What you need before you install

Check these requirements first. The package is standard-library Python only, so it needs no dependency installation, but it does need a network connection and a paid account for renders.

• Python 3.10 or newer.
• git, used to fetch the pinned package tree from GitHub.
• A Beatra account, with a browser sign-in to authorize the skill.
• A paid credit balance for renders.
• Network access to github.com and mcp.beatra.ai.
• Willingness to review the bundled Python scripts before running them.

Install the skill from a reviewed, pinned copy

Download the ZIP from the directory page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. In the Claude apps you can instead upload it under Settings, then Capabilities, then Skills.

The source's own install method fetches a fixed commit and checks its tree ID before copying anything. Ask for approval before the first step, and stop if the last command does not print TREE-OK. Then inspect the files, and only after a second, separate approval copy them in. Read every file in the pinned tree yourself. Grep output is only an inventory aid, not proof of safety.

• Step 1 fetches the pinned tree into a temporary review directory.
• Step 2 lists any symlink, executable, or submodule entries and checks for self-updater code.
• Step 3 copies the reviewed folder into ~/.claude/skills/ai-logo-maker after confirming the destination does not already exist.
• Step 4 runs authorize.py, which opens a browser sign-in and stores a bearer token. Ask before running it.

Trigger the skill with natural requests

You do not need special syntax. Ask for logo work in plain language, and say clearly that you accept a paid render on Beatra. The skill is meant for users who explicitly want the work done there and agree to send prompts and any reference images to a third party. If you want local-only editing or no paid render, do not invoke it.

Start a new agent session if your host only discovers skills at startup.

• Make me three logo concepts for a coffee roastery called Ember & Oak, using Beatra, I accept the paid render.
• Turn this sketch into a vector-friendly logo mark on Beatra. I approve sending the file.
• Check whether this logo will stay legible at 16 pixels before we use it as a favicon.

Approve each paid call

Before any paid call, the skill follows a fixed sequence. It first uses free discovery to read current model cards and credit estimates. It then shows you a cost card for exactly one paid call, listing the tool, model, estimate, and which files will be uploaded. It waits for explicit approval. Planning or a phrase like just do it does not count as approval.

After a render, the skill reports the delivered artifact and the billing.net_charged_credits value, then asks again before any further paid stage.

• Only files you supplied and approved for this task are uploaded.
• Each paid submission uses a stable client_request_id, so a lost response can be retried without creating a second job.
• Run the free connection check first to confirm the package can reach the service.

Safety notes you should read first

This skill runs scripts, makes network calls, and stores a bearer token. Those are the facts that matter most, so review them before you activate it. The package has no self-updater that downloads code or changes its own files, and this version has no switch to turn off updates because it has no update code.

Do not use it for confidential brand material that must stay inside your organisation, because prompts and uploaded images go to a third-party service. Do not run it if you cannot review its Python scripts.

• It runs three Python scripts: authorize.py, mcp_client.py, and uninstall.py.
• It stores credentials and local state under ~/.beatra.
• The update command only reports what the installing store offers. For a copy installed by these steps it finds no record and tells you to update the way you installed it.
• To move to a newer version, repeat the review steps for that commit in a new directory, then replace the folder only after approval.