Die Oberfläche ist lokalisiert; Skill-Beschreibungen und -Dokumente werden per KI automatisch in Ihre Sprache übersetzt. Code, Befehle, Pfade und Fachbegriffe bleiben im Original, maßgeblich ist das Dokument des Autors.
Skills entdecken
Produktivität · SKILL.md

bank-desk-board-set

bank-desk-board-set

Turn user-supplied branch window names and board lines into a four-to-eight still bank desk board set. This window-to-board studio lays out each named window as its own branch board still. Use it for branch window board packs, teller board stills, and matching desk board graphics.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill-Dokumentation

GitHub

name: "bank-desk-board-set" description: "Turn user-supplied branch window names and board lines into a four-to-eight still bank desk board set. This window-to-board studio lays out each named window as its own branch board still. Use it for branch window board packs, teller board stills, and matching desk board graphics."

Bank Desk Board Pack

Make a four-to-eight still pack from window names and board lines the branch already supplied. Use this Skill when a hall window needs a matching branch board pack — one still per named window — not a resident-event pack and not a session cover pack.

Use resident-event-set when the work is one still per named activity. Use wealth-cover-set when the work is one still per named investor session. Use hanzi-card-set when the work is one still per named character. Do not look up public posts for missing window titles. Do not invent rates, hours, a ranking, an official stamp, or a red-head notice.

Collect supplied windows

Hard inputs are the exact pack name and the user-supplied window names and board lines that will appear on the stills. Do not invent a window, rate, hour, ranking, or official stamp to finish a slot. Reuse destination (print, screen, slide), language, palette, and must-keeps already in the conversation.

Ask only for facts that change the pack:

  • exact branch, hall, or pack name;
  • each named window and the board lines the branch already approved;
  • destination surface and language; and
  • an optional brand still if the pack should match that look.

A scan or photo is a visual reference, not a source for missing window titles. Uploading makes media available to Beatra and does not inspect it. When the host cannot view a file, keep the user's declared role for it.

Default to one still per named window, four to eight stills in the pack. If they name fewer than four windows, keep that count. Cap the pack at eight. If they want a standard set and have not numbered windows, plan four stills from the names already given. Omit a line whose fact is missing and keep that gap on the plan so the branch can add it later.

Keep the same type, palette, and layout across the pack so the stills read as one branch board pack.

Plan the free pack list

Write a labeled pack list before any paid image. For each window record the printed lines, language, layout (window name, board points, optional footer), canvas, and any optional reference role. That list is the free visible result. Planning is not approval.

Before setting a concrete model, canvas, output count, or price, read the current beatra.models.list card for text_to_image. Keep model as auto and count at 1 per still. Default canvas is 3:4 2K for print hall boards. Change canvas only when the user names a destination that needs another ratio. A user who wants another candidate adds a distinct slot and approves that new work.

Use beatra.images.edit only after the user accepts a still and asks for a local correction, with that accepted image as images[0].

Read bank desk board pack workflow when writing the payload or recovering a task.

Confirm once, create by window

Planning is free. Before the first billable call, show one current production card that lists every named window still, and wait:

  1. Work — one window still per named desk, a generate call (beatra.images.generate).
  2. Credits — the live text_to_image price just read. Do not reuse a remembered number.
  3. Count — one paid call per window still. This card lists every slot in the pack. Do not batch several windows into one count. This approval does not cover retries.
  4. Identity — one new opaque client_request_id per still. A changed prompt, window line, file, model, or canvas mints a new ID.
  5. If we stop here — the labeled pack list remains usable.
  6. If the balance is insufficient — relay the official message and its top-up URL exactly. Translate the prose; keep the URL. Do not retry until the user says they have topped up.

Submit once through bundled scripts/mcp_client.py. Poll beatra.tasks.get. Deliver actual bytes plus billing.net_charged_credits. Do not promise the prepaid estimate is the final charge.

After approval, submit each still once. Keep no more than two generation tasks in flight on one connection.

Review, deliver, and recover

Review printed lines against the confirmed pack list. Report only the text the host can actually see. Treat generated small type as a review item, not as an official notice, a rate board, or a published hour.

Deliver the stills in window order, the pack list, observed dimensions and formats, task IDs, resolved models, and returned billing.net_charged_credits. A focused correction is new paid work with its own card and ID.

After a returned task_id, poll that task. If the create response is lost, retry only the identical frozen payload with the original ID. If the task ID is missing, use beatra.tasks.list and verify candidates with beatra.tasks.get before another submit. Use beatra.tasks.cancel only when the user asks.

Execution

Invoke every remote Beatra operation only through this package's bundled scripts/mcp_client.py. Put the MCP tool name after call and send one JSON object on standard input.

python3 scripts/mcp_client.py call beatra.models.list
{"capability": "text_to_image"}
printf '%s' '{"prompt":"Create the approved bank-desk board still for the named window. Print only the confirmed window name and board lines. Keep the shared pack look.","model":"auto","count":1,"canvas":{"type":"preset","tier":"2K","aspect":"3:4"},"client_request_id":"opaque-bank-desk-board-01"}' | python3 scripts/mcp_client.py call beatra.images.generate

Do not configure or call a host Beatra Connector, and do not use REST/OpenAPI as a fallback.

If the tool returns insufficient credits: give the user the tool message in full. Do not change the URL. You may translate the other words. Do not replace it with your own top-up copy. Point the user to pay using that message.

Account balance

When the user asks how many credits remain or whether a live estimate fits, call beatra.wallet.get. When they ask what was charged, call beatra.wallet.ledger. Both are read-only. Do not invent an account-balance or top-up tool. Do not make wallet.get a required step before every paid submit.

When a model card comes back carrying a top_up block, relay its tiers as the card lists them and in that order. Do not rank them, do not talk one down, and do not pick one for the user. Which tier suits them is their call, made on the wallet page with the whole list in front of them. Never quote a tier from memory.

References by task

Runtime and safe automatic updates

The bundled client silently checks for a newer release at most once every 24 hours per installation. When a newer version is available, it installs automatically without separate confirmation. It downloads only from the fixed official Beatra discovery and immutable CDN paths for this package, channel, and locale, verifies discovery data, archive, manifest, and every packaged file, and replaces only package-owned files.

Update checks, downloads, verification, replacement, rollback, and recovery fail open: the current installation remains usable and the original command continues. An update failure never authorizes retrying a paid image request. The setting persists for this installation. See automatic updates and safety.

python3 scripts/mcp_client.py update --auto off
python3 scripts/mcp_client.py update --auto on
python3 scripts/mcp_client.py update --check
Gespeicherte Skills