School Song Maker

Turn a school occasion into original lyrics and a finished song a class can sing together: a class song, a graduation song, a freshman welcome song, a new school or college anthem, a teachers' day song, a school anniversary song, or a sports day or military training song.

What you get

• One lyric sheet — title, labeled sections (verse, chorus, bridge), the full lyrics, every must-include line marked where it is sung, and singability notes for the group that will sing it.
• One finished song from that lyric sheet, sung in the chosen language with the chosen voice (student choir, children's choir, or a solo voice). The model may return more than one take; deliver every take it returns. Voice and length are guided by the prompt (about two and a half to three minutes); the actual voice and duration are reported on delivery.
• Optional: one instrumental piece for an entrance, an award moment, or background music at the same event.

Ways to run it

• One song (default) — lyric sheet, then one song.
• Song plus instrumental — the same lyric sheet and song, plus one instrumental piece for the event. Both slots appear on one production card, each with its own request identity. The instrumental is a new piece in the same style, not the delivered song without vocals: a song cannot be split into vocals and accompaniment here. When the user asks for a backing track to sing the song live, say so and write it on the card's Work line. No sheet music or score is produced.
• One song per class — the same occasion for several classes (for example the welcome song for classes 1 to 4 of one year group). Make the first class's song first; the remaining classes become a labeled set after that first song is accepted.
• Lyrics only — stop after the lyric sheet when the class will set or sing the words themselves.

Scope and adjacent routes

Use it when a teacher, class leader, student union, or school office wants a song written for a school occasion from the school's own facts. Route a finished lyric that only needs a song to suno-lyrics-to-song; a new arrangement of an existing recorded school anthem to ai-song-cover-studio; a personal gift song for one person to personalized-song-maker; an exam knowledge-point song to exam-mnemonic-song; and spoken lecture narration to course-narration-studio. Route mixed or uncertain music requests to beatra-ai-music-creator.

Inputs and defaults

Hard inputs:

• the occasion (class song, graduation, freshman welcome, school anthem, teachers' day, anniversary, sports day, military training, or another named school event);
• who sings it: the school stage (primary, middle, high school, vocational, or college) and whether a whole class sings along;
• the identity to sing: school name, class or major, and year, with the exact form the user wants sung (full name or short name);
• must-include lines: motto, class slogan, names, or memories the user supplies;
• the language to sing.

Ask only for a missing hard input. Reuse anything already given. Do not invent a motto, a school history fact, a memory, or a student name to finish a lyric. When the user lists student or teacher names, sing only the names they list, written exactly as they gave them.

Safe defaults:

• Mood follows the occasion: bright and forward-looking for welcome and sports songs, warm and grateful for graduation and teachers' day, steady and proud for an anthem or anniversary.
• Voice follows the singers: a children's choir for primary school, a mixed student choir singing the chorus in unison for a whole class, a solo voice only when the user asks for one.
• Melody serves a group: short lines, a repeated chorus with one clear hook, a comfortable middle range, and a moderate tempo.
• model: "suno-6" with model_options.suno.custom_mode: true, so the approved lyrics are sung as written. Show the model on the production card; switch only when the user chooses another live model.
• instrumental: false with the approved lyrics present. Style lives only in prompt; the sung words live only in lyrics. There is no duration field; write the target length in the prompt.

Golden path

1. Collect the intake. Record occasion, singers, identity to sing, must-include lines, memories, language, and mood.
2. Write the free lyric sheet. Title, labeled sections, full lyrics, each must-include line mapped to where it is sung, and singability notes. Add the production prompt: genre, mood, tempo, instrumentation, and voice. This sheet is the free visible result; revise it with the user until they are happy. Planning is not approval.
3. Read the live model card. Call beatra.models.list for text_to_music and confirm the live prompt, lyrics, and title limits, the controls, and the current price.
4. Show the production card and wait (next section).
5. Submit once through the bundled client (see Execution) with one opaque client_request_id per slot. Record each returned task_id immediately.
6. Poll and deliver. Poll beatra.tasks.get until terminal. Deliver every returned take with its link, actual duration, resolved model, and billing.net_charged_credits. Review it (see Review, deliver, and recover).

Decisions that require confirmation

Music generation is the only paid step. Before any billable call, show one current production card and wait for a clear yes:

1. Work — what will be made in the user's words (for example "the 2026 class 2 welcome song" and, when chosen, "one instrumental entrance piece, a new piece in the same style"), each a beatra.music.generate call.
2. Credits — the live text_to_music price just read from beatra.models.list, per slot and in total. Never a remembered number.
3. Count — one paid call per slot listed on the card. A later class, a changed lyric, or a second try is another slot.
4. Identity — one new opaque client_request_id per slot. Any changed lyric, title, prompt, model, or control mints a new ID.
5. If we stop here — the lyric sheet stays complete and 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.

One approval covers only the slots on that card. In the one-song-per-class mode, accepting the first song is not approval of the rest: show a fresh card for the remaining classes.

Execution

Invoke every remote Beatra operation only through the 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_music"}

Send the generate payload through a quoted heredoc (or a JSON file on standard input), never inside a single-quoted shell string: lyrics and school names often contain apostrophes, and a single-quoted string silently drops them. The lyrics and title sent must match the approved sheet character for character; inside the JSON, escape double quotes, backslashes, and control characters such as line breaks (\n) and tabs (\t).

python3 scripts/mcp_client.py call beatra.music.generate <<'JSON'
{"model":"suno-6","prompt":"Bright campus pop anthem, mixed student choir singing the chorus in unison, piano and strings with light drums, moderate tempo, easy to sing along, about two and a half minutes.","lyrics":"[Verse 1]\nWe're the class that never quits\n[Chorus]\nApproved hook","instrumental":false,"title":"Approved Title","model_options":{"suno":{"custom_mode":true}},"client_request_id":"opaque-school-song-01"}
JSON

Do not configure or call a host Beatra Connector, and do not use REST/OpenAPI as a fallback. Exact payloads for the instrumental slot and the per-class set are in the workflow reference.

Review, deliver, and recover

Review each take against the lyric sheet: the school and class names sung in the form the user chose, every must-include line present, the chorus clear enough for a class to pick up, the voice matching the chosen choir or solo voice, the mood right for the occasion, and the audio clean. Report only what the host could actually hear; if it could not listen, say so and list what the user should check.

If the user wants changes after a delivered song, a changed lyric, title, prompt, or model is new paid work: revise the sheet for free, then show a new card with a new ID.

If a create response is lost, call beatra.tasks.list for text_to_music and match candidates with beatra.tasks.get against the private ledger before any replay; reuse an ID only with byte-identical arguments. Queued and running are progress states. Call beatra.tasks.cancel only when the user asks, then confirm the terminal state.

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, and do not make wallet.get a required step before every paid submit.

References by task

• Read School song workflow when collecting the intake, writing the lyric sheet, choosing the voice for an age group, building payloads for the instrumental slot or the per-class set, recovering a task, or answering a request for a backing track or a score.
• Read Installation and authentication only when authorization or shared credentials need attention, and Installation registration for the non-billable registration step.
• Read Tasks and results for terminal task and artifact semantics, and Billing, errors, and recovery for returned billing or error details.
• Read Bundled MCP Client diagnostics when the bundled client cannot connect. Do not configure a host Connector.
• Read Automatic updates and safety for update guarantees and controls, and Uninstall and disconnect only when the user asks to stop using it and remove shared credentials.

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 music 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