How to use the AI Headshot Studio agent skill

What the AI Headshot Studio skill makes the agent do

An agent skill is a folder of instructions that an AI agent reads when it is installed, so the agent knows how to carry out a specific task. This one covers professional headshots for resumes, company websites, career profiles and social media.

The catalog entry is a pointer to an external package. It contains no image-editing 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.

• Turns a casual selfie into a headshot with a new background, business attire and studio lighting.
• Works from inside Claude Code, Codex or OpenClaw, according to the skill's overview.
• Produces the work on the hosted Beatra service at mcp.beatra.ai, which is paid.

What you need before installing

Check these requirements first. The skill will not run without a Beatra account and paid credits, and the install steps use git and Python.

• Python 3.10 or newer. The package uses only the standard library and installs no dependencies.
• git, which the installer uses to fetch the pinned package.
• A Beatra account, with browser sign-in to authorize the service.
• Paid credits on the Beatra service for rendering.
• Permission to send your photos and prompts to an outside service.

Install the skill from the pinned package

The skill's install method fetches one specific commit and checks its tree ID, so you know exactly which files you are reviewing. You can also get the skill as a ZIP from this directory page and unzip it into ~/.claude/skills/ for your user, or into a project's .claude/skills/ folder. In the Claude apps you can upload it under Settings, then Capabilities, then Skills.

Whichever route you use, the source says to read every file in the package before activation, including each script, reference and hook. Grep and manifest checks only show what is there; they are not a review. If you cannot complete the review, stop and do not copy or run the files.

The first step clones the repository without checkouts and then checks out only the skill folder at the pinned commit. The command stops unless the tree ID matches.

• Confirm that the output ends with TREE-OK before going further.
• The catalog lists the license as MIT; the skill's own file header says MIT-0 (MIT No Attribution).
• The package has no self-updater and never downloads code or changes its own files, according to the source.

Copy the reviewed skill into your skills folder and authorize Beatra

After you have read every file, copy the reviewed folder into your Claude Code skills directory. The copy stops if the destination already exists, so you will not overwrite an earlier install by accident. Then authorize Beatra. Authorization opens a browser sign-in and stores a bearer token, so the agent asks before running it.

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

• The copy goes to ~/.claude/skills/ai-headshot-studio.
• The authorize script is scripts/authorize.py inside that folder.
• Other hosts use their equivalent skills path; the source shows Claude Code.

How to ask for a headshot and approve the cost

Ask in plain language that names the result and says you accept a paid render. The skill is meant for requests that explicitly want Beatra headshot work and accept the charge. It does not apply to local-only image editing.

Before any paid call, the agent shows a cost card with the tool, the model, the estimate and the files to be uploaded. It waits for your explicit approval. It uploads only files you supplied for the task, submits once, and reports the billing figure for the net charge. Approval covers the named service, file set, exact call and cost ceiling; an open-ended request does not approve unknown charges.

• Example request: 'Make a professional headshot from this selfie for my resume. I accept the paid Beatra render.'
• Example request: 'Show me the cost before you upload anything, then make a company-website headshot from this photo.'
• Example check: 'Run the free connection check and tell me whether Beatra is reachable.' This is non-billable.

Security and safety notes to read before you approve

This skill sends images and prompts off your device to a third party, and it stores a credential on your machine. The source states that the scripts make network calls and store a bearer token. Treat that as a real privacy and security decision.

If a retry is needed after a lost response, the agent should reuse the same request id and identical arguments rather than submit a replacement while a task is running. Ask the agent to show you the details before any further paid stage.

• Only share photos you are comfortable sending to Beatra.
• Skipping the file review is a security risk; do not skip it.
• There is no offline option for this skill.
• Keep the install review directory until you have finished the review.

Update or uninstall the skill

This pinned version cannot update itself. To move to a newer version, repeat the fetch and review steps for that commit in a new review directory, compare the changes with the installed copy, and replace the folder only after you approve. A version not listed in the catalog entry has not been reviewed.

To uninstall, run the dry run first and read the JSON decision. If the output says disconnected, this was the last Beatra skill on the device, and the token is revoked when the output marks revoked as true. The script then removes the local Beatra credential and state files under ~/.beatra.

• Run the dry run first to see the decision without changes.
• A disconnected result means local Beatra credentials and state files are removed.