Hanzi Card Set

Make a four-to-eight still pack from a character list the user already supplied. Use this Skill when a lesson needs a hanzi card set — one still per named character or named card group — not an art demo page and not a product gallery.

Use art-demo-set when the work is one classroom still per drawing step. Use allergen-label-card for a seller-fact still. Keep risk grade boards for risk-grade-set when that package is installed. Do not look up public posts, textbooks, or dictionaries for missing characters. Do not invent a score, rank, or grade for the student.

Collect supplied characters

Hard inputs are the exact set name and the user-supplied character list that will appear on the cards. Do not invent a character, pinyin, gloss, stroke count, example word, score, or rank to finish a slot. Reuse destination (handout, board, screen), language, palette, and must-keeps already in the conversation.

Ask only for facts that change the pack:

• exact lesson, unit, or set name;
• each named character or named card group the user already approved;
• any pinyin, gloss, or example word the user already wants printed;
• destination surface and language; and
• an optional brand or sample still if the pack should match that look.

A scan or photo is a visual reference, not a source for missing characters. 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 character, four to eight stills in the pack. If they name fewer than four characters, keep that count. Cap the pack at eight. If they group characters onto named cards, honor those groups instead of splitting them. If they want a standard set and have not numbered cards, plan four stills from the characters already given. Omit a line whose fact is missing and keep that gap on the plan so the user can add it later. Extra characters beyond eight stay as named gaps, not a ninth still.

Keep the same type, palette, and layout across the pack so the stills read as one hanzi card set.

Plan the free pack list

Write a labeled pack list before any paid image. For each card record the printed lines, language, layout (character, optional pinyin or gloss, 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 recognition cards. 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 hanzi card pack workflow when writing the payload or recovering a task.

Confirm once, create by card

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

1. Work — one recognition-card still per named character or named card group, 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 card still. This card lists every slot in the pack. Do not batch several cards into one count. This approval does not cover retries.
4. Identity — one new opaque client_request_id per still. A changed prompt, character 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 a scored sample or a certified reading level.

Deliver the stills in list 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 hanzi recognition-card still for the named character. Print only the confirmed character and supplied lines. Keep the shared pack look.","model":"auto","count":1,"canvas":{"type":"preset","tier":"2K","aspect":"3:4"},"client_request_id":"opaque-hanzi-card-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.

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

• For pack lists, payloads, confirmation, polling, recovery, and review, read hanzi card pack workflow.
• For authorization and the non-billable registration step, read installation and authentication and installation registration.
• For shared task, billing, and connection details, read tasks and results, billing, errors, and recovery, and Bundled MCP Client diagnostics.
• For update guarantees and controls, read automatic updates and safety. For removal, read uninstall and disconnect.

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