suno-music · calesthio — скилл для Claude Code: что делает и как установить
Скилл suno-music для Claude Code создаёт музыку в Suno и разбирает права на её использование. Он включается для генерации, редактирования треков и подготовки промптов.
MIT 157 установок ★ 252 обновлён 2026-07-14
npx skills add calesthio/generative-media-skills --skill suno-music -a claude-codeЧто делает suno-music
Скилл suno-music помогает создавать в Suno песни с вокалом и инструментами, продлевать, ремастерить и разделять треки на стемы. Он формирует промпты для полей style и lyrics, структурных метатегов, exclude-styles, personas и custom voices, а также объясняет выбор подписки.
Отдельная часть отвечает за владение результатом, коммерческие права, выпуск и монетизацию на DSP, соглашение с Warner Music Group, судебные споры и доступность API Suno. Для коммерческой сделки с конкретными условиями скилл не заменяет юриста. V5 доступна подписчикам Pro и Premier, а бесплатный тариф работает со старыми моделями.
Какие задачи решает
- создать трек в Suno по жанру, настроению и инструментам
- продлить, ремастерить или разделить песню на стемы
- составить промпт для style, lyrics и exclude-styles
- разобраться с коммерческими правами и выпуском трека на DSP
Примеры запросов агенту
- Создай в Suno трек в стиле
melodic drum-and-bassс женским альтом и синтезаторным басом. - Продли этот трек, добавь метатеги
[Chorus]и[Outro], а нежелательные стили укажи вexclude-styles. - Объясни, какие права нужны для коммерческого выпуска результата Suno и какой тариф выбрать для V5.
Как установить suno-music
1Через Skills CLI (Vercel)
npx skills add calesthio/generative-media-skills --skill suno-music -a claude-code2Вручную: положить файлы
~/.claude/skills/suno-music/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/suno-music/
1Через Skills CLI (Vercel)
npx skills add calesthio/generative-media-skills --skill suno-music -a codex2Вручную: положить файлы
.agents/skills/suno-music/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add calesthio/generative-media-skills --skill suno-music -a cursor2Вручную: положить файлы
.cursor/skills/suno-music/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install calesthio/generative-media-skills skills/providers/music-generation/suno-music2Вручную: положить файлы
.github/skills/suno-music/SKILL.md1Через Skills CLI (Vercel)
npx skills add calesthio/generative-media-skills --skill suno-music -a gemini-cli2Командой платформы
gemini skills install https://github.com/calesthio/generative-media-skills --path skills/providers/music-generation/suno-musicВручную: положить файлы
.devin/skills/suno-music/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Через Skills CLI (Vercel)
npx skills add calesthio/generative-media-skills --skill suno-music -a opencodeClaude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о suno-music
Как установить suno-music в Claude Code?
Выполни npx skills add calesthio/generative-media-skills --skill suno-music -a claude-code, затем используй файл ~/.claude/skills/suno-music/SKILL.md.
Что делает suno-music?
Скилл помогает генерировать, продлевать, ремастерить и разделять музыку в Suno. Он также готовит промпты и объясняет вопросы владения, коммерческого использования, дистрибуции и API.
Что нужно для работы suno-music?
Нужен доступ к Suno. Для V5 требуется платная подписка Pro или Premier; бесплатный тариф использует более старые модели.
Исходный текст
SKILL.md · MIT
--- name: suno-music description: >- Generate music with Suno (v5 / v5.5 era, 2026) and advise a production team on what they may legally do with the output. Use this skill when a user wants to create, extend, remaster, or stem-separate a track in Suno; craft Suno prompts (style field, lyrics/metatags, exclude-styles, personas, custom voices); choose a subscription tier for a given use; or get accurate answers about ownership, commercial rights, distribution/monetization on DSPs, the Warner Music Group settlement, the ongoing label/publisher litigation, and Suno's API situation. Do not use it for other music generators (Udio, ElevenLabs Music, Google Lyria) except by explicit contrast, and do not use it as a substitute for a lawyer on a specific commercial deal. --- # Suno music generation and rights advisory Suno is a text-to-music service that generates full songs (vocals + instrumental) from a style description and optional lyrics, plus editing tools (extend, cover, remaster, stem separation, a DAW-style Studio). This skill covers two things a production agent must get right at the same time: **making good music with the current models** and **advising accurately on the rights, licensing, and legal reality of using that music commercially**. The rights half is not an appendix. As of mid-2026 the single most common way to give a production team bad advice is to treat Suno output as freely ownable, copyrightable, and API-accessible. Read the "Rights, licensing, and legal reality" section before advising on any commercial or distribution use. **Evidence labels used below** - **[Documented]** — stated in Suno's own docs, terms, help center, or blog, or in a primary legal/press source, with a URL in Sources. - **[First-party claim]** — Suno's own marketing/announcement language, reported as a claim rather than an independently verified fact. - **[Heuristic]** — production practice inferred from how the tools behave; useful defaults, not guarantees. **Volatility warning.** Models, tiers, prices, terms, litigation status, and API access change fast. Every volatile fact below carries a verification date of **2026-07-10** unless noted. Re-verify anything legal or version-specific against suno.com before relying on it. --- ## 1. Current models and what they do **[Documented] Model timeline** (Suno help center "Model Timeline", verified 2026-07-10): | Model | Released | What it added | |-------|----------|---------------| | V3.5 | Summer 2024 | Better structure; up to ~4 min first gen | | V4 | Nov 2024 | Improved vocal quality; introduced Extend, Cover, Persona | | V4.5 | May 2025 | Up to 8-min first generation; better prompt adherence; style mashups | | V4.5+ | Jul 2025 | Add Vocals / Add Instrumental production tools | | V5 | Sep 2025 | Higher audio fidelity, more authentic vocals; **Pro/Premier only** | | V5.5 | Mar 26 2026 | Voices (record/upload your own voice), Custom Models, My Taste | - **[Documented]** V5 launched ~September 23–25, 2025 and is available only to paid (Pro/Premier) subscribers; free-tier users generate on older models. Suno also shipped **Suno Studio** (a generative audio workstation with a multitrack editor and MIDI/stem export) on the Premier plan around Sep 25, 2025. - **[Documented]** V5.5 (Mar 26 2026) added three features: **Voices** (capture or upload your own voice, sing on your creations; includes voice verification, recordings kept private), **Custom Models** (Pro/Premier can train up to three personal model versions on their own catalog), and **My Taste** (preference personalization, available to all tiers). - **[First-party claim]** Suno marketed V5 as its "most powerful" / "world's best music model" and V5.5 as "more expressive." Treat superlatives as marketing, not benchmarks — there is no neutral, published head-to-head that establishes Suno as a universal "best." When a user asks "is Suno the best," answer on fitness for their use, not on the marketing claim. - **[Documented — anticipated]** Under the Warner deal (see §5), Suno has said higher-quality models **trained on licensed WMG recordings** are forthcoming. As of 2026-07-10 no such licensed model is confirmed shipped under a public name; treat "WMG-licensed model" as announced-but-unshipped and re-verify. **Version selection heuristic.** For any track intended for release, generate on V5/V5.5 (paid). Free-tier output is both lower quality *and* non-commercial (§4), so it is only appropriate for scratch ideas, learning, and personal listening. --- ## 2. The generation surface: fields and controls Suno's create screen (Custom mode) has two text inputs that do different jobs — mixing them up is the most common cause of bad output. - **Style / description field** — genre, subgenre, mood, tempo feel, instrumentation, production character, vocal type, era, reference *adjectives*. This is where the sonic identity lives. - **Lyrics field** — the actual words **plus structural metatags in square brackets**. If you put genre words here they may be sung. If you leave it empty (or use "Instrumental"), you get an instrumental. - **[Documented] Exclude Styles** (negative prompt) — in Custom mode under Advanced Options; list what you do *not* want (e.g. a genre, an instrument, "male vocals") and Suno deprioritizes it. - **Simple vs Custom mode** — Simple mode takes one free-text prompt and writes lyrics/structure for you; Custom mode exposes the style field, lyrics field, exclude-styles, model picker, and weirdness/style-influence sliders. Use Custom for anything you want control over. **[Heuristic] Style field craft** - Lead with the anchor: one clear genre + one clear mood, then layer detail. `melodic drum-and-bass, euphoric, driving` beats a 40-word soup of adjectives. - Name instruments and vocal type explicitly when they matter (`female alto, breathy`, `analog synth bass`, `brushed jazz drums`). - Tempo: you can state BPM in the style field, or use a `[bpm-128]` metatag in lyrics; giving a numeric BPM is more reliable than "fast." - Don't over-stuff. Beyond ~6–10 strong descriptors, added words dilute rather than sharpen. - Contradictions (e.g. "minimal, wall-of-sound maximalist") produce mush; pick a lane. --- ## 3. Lyrics, structure, and metatags **[Documented]** Structural **metatags** go in the *lyrics* field inside square brackets and steer arrangement, dynamics, and vocal behavior. Without them Suno infers structure from line breaks; with them you remove the guesswork. Common, reliably-understood tags (**[Documented]** for the core structure tags; **[Heuristic]** for exact behavior, which varies by model): ``` [Intro] [Verse] [Verse 1] [Verse 2] [Pre-Chorus] [Chorus] [Post-Chorus] [Bridge] [Instrumental] [Guitar Solo] [Drop] [Outro] [Fade Out] [no-vocals] [bpm-128] [key-C-minor] [time-signature-3/4] ``` **[Heuristic] Lyric and structure practice** - `[Chorus]` triggers chorus behavior — bigger arrangement, hook, repetition on later occurrences. Reusing the exact same chorus text each time yields a consistent, memorable hook; varying it slightly yields evolution. - Keep tags short (1–3 words). Long descriptive tags are less reliable than a short structural tag plus a style-field descriptor. - To force an instrumental, put `[Instrumental]` / `[no-vocals]` and leave lyrics empty, or select the Instrumental toggle. - Inline performance cues (whispered, ad-lib, spoken) work inconsistently across models — test, don't assume. - Suno can auto-write lyrics; for a controlled brief, write your own lyrics so you also own that copyright layer (§4). --- ## 4. Editing, iteration, and stems - **Extend** — continue a track past its current end; useful for building full arrangements from a strong 1–2 min section. - **Cover** — re-perform existing lyrics/melody in a new style. - **Remaster** — **[Documented]** re-render an older track on a newer model (e.g. lift a V4 track to V5 quality), with control over how much it varies from the original. - **Add Vocals / Add Instrumental** — layer onto an existing track (V4.5+ onward). - **Personas / Voices** — **[Documented]** Persona captures the vocal/stylistic identity of a prior generation to carry it across new songs; V5.5 **Voices** lets you record or upload your *own* voice to sing on creations (with verification; recordings private by default). Do not use a Voice/Persona to imitate a real artist you have no rights to (§5, §6). - **Stems** — **[Documented / First-party claim]** Suno can split a generation into individual stems (vocals, drums, bass, and more — marketed as up to ~12 stems) and Suno Studio (Premier) exports multitrack stems and MIDI for finishing in a DAW (Ableton, Logic, FL, etc.). **[Heuristic] Iteration workflow.** Generate several candidates on a tight style prompt → pick the best structural take → Extend to full length → Remaster/clean if needed → export stems for mix and master in a DAW. Regenerating from a slightly varied prompt is usually faster than fighting one bad take with edits. --- ## 5. Rights, licensing, and legal reality (read before advising on commercial use) This section is the core of the skill. Get the tier logic and the copyright caveat right every time. ### 5.1 Ownership and commercial use by tier **[Documented]** (Suno Terms of Service, last revised **2026-03-26**; Suno help center "Rights & Ownership"; verified 2026-07-10): - **Free / Basic tier:** Suno **retains ownership**. You get a limited license to use output for **lawful, internal, personal, non-commercial** purposes **only**, and you must **give attribution to Suno**. No distribution, no monetization, no commercial use. - **Pro / Premier (paid) tiers:** Suno **assigns to you** all of its right, title, and interest in the output generated from your submissions **while you were subscribed**. You may distribute, monetize, and use it commercially. Premier gets the same commercial *rights* as Pro — the extra cost buys volume (credits) and tooling (Studio), **not** broader legal rights. - **Timing is what matters, not current status:** ownership/commercial rights attach based on your subscription **at the moment the track was generated**. **[Documented]** Songs made while paid stay commercial-usable **even after you cancel**. Conversely, upgrading later does **not** retroactively grant rights to a track you made on the free tier — you must regenerate it while paid. **Approximate 2026 pricing (verify — volatile):** Pro ≈ $10/mo (~$8 annual), ~2,500 credits; Premier ≈ $30/mo (~$24 annual), ~10,000 credits + Suno Studio. Prices/credits change; confirm on suno.com before quoting. ### 5.2 The copyright caveat that ownership does not cure **[Documented]** Two separate things, and agents routinely conflate them: 1. **Contractual ownership from Suno** — paid tiers get this (Suno assigns its interest to you). 2. **Copyright protection under law** — a *different* question. Suno's terms state it makes **no representation or warranty that any copyright will vest in any Output**. Suno's own help center warns output "may not be eligible for copyright protection." **Why:** U.S. copyright requires human authorship; the U.S. Copyright Office has repeatedly held that material generated by AI from a prompt is **not** protectable, because writing a prompt is not authorship of the resulting expression. **[Documented]** Practical consequence: a fully Suno-generated track can be **owned** (contractually, on a paid tier) and **legally used commercially**, yet **not copyrightable**, meaning you may have limited ability to stop others from copying it. **[Heuristic]** The strongest human-authorship hook is **writing your own lyrics** (you own that literary copyright independently) and doing substantial human arrangement/editing; this improves — but does not guarantee — registrability. This varies by country; direct consequential decisions to a qualified attorney or the relevant copyright office. ### 5.3 Distribution, monetization, and disclosure - **[Documented]** Suno officially says paid-tier songs can be distributed to Spotify, Apple Music, etc.; free-tier songs cannot be monetized. - **[Documented]** Distributors (e.g. DistroKid) allow AI music in 2026 but require proof of commercial rights — you must have been on a **paid tier when the audio was generated** — and you must be the exclusive holder of 100% of the material (don't use someone else's lyrics/melody). - **[Documented] AI disclosure:** From late 2025, DSPs began enforcing the **DDEX** industry standard for AI-disclosure metadata; a track using Suno-generated audio should be flagged as AI-generated during distributor upload. This is a contractual/labeling requirement, not optional. - **[Documented] Content ID:** As of mid-2025, fully AI-generated audio is **not eligible** for YouTube Content ID. You can monetize your *own* videos using the track, but you cannot claim Content ID to monetize third-party uses. Do not promise a client Content ID revenue on pure Suno audio. ### 5.4 The Warner Music Group settlement and the licensing landscape - **[Documented]** On **2025-11-25**, Warner Music Group and Suno announced a **settlement of WMG's copyright suit plus a licensing partnership**. Suno gets licensed access to WMG catalog as source material for "next-generation licensed AI music"; artists/songwriters can **opt in** to features that let users create content using their name, image, likeness, voice, and compositions. Suno also **acquired Songkick** (concert discovery) from WMG. **Financial terms were not disclosed.** - **[Documented — anticipated]** Suno signaled 2026 changes under the deal: models trained on licensed WMG recordings are forthcoming, "you'll still be able to create original songs the way you do today," and download access remains monthly-limited for paid users while **Suno Studio users keep unlimited downloads**. Treat specifics as forward-looking; re-verify. - **[Documented]** Udio settled with **Universal Music Group** on **2025-10-29** (payment + license for a joint AI platform). So by mid-2026, of the three majors: **UMG ↔ Udio** and **WMG ↔ Suno** have deals. ### 5.5 Litigation that is still open (as of 2026-07-10) - **[Documented]** **Sony Music has NOT settled** with Suno (or Udio). Sony's fair-use cases are expected to produce a pivotal ruling around **summer 2026** that could set precedent for the whole sector. **Do not tell a client the legal question is settled.** - **[Documented]** **Suno is contesting** on fair-use grounds — it filed a motion for summary judgment in **March 2026** arguing that training on copyrighted music is transformative use under 17 U.S.C. §107. Reported **UMG–Suno settlement talks hit an impasse** (April 2026). - **[Documented]** Additional suits: the **American Federation of Musicians (AFM)** sued UMG and WMG over how the Suno/Udio settlements were structured, and **independent musicians filed class actions** arguing the major-label deals don't protect smaller rights holders. - **Advisory takeaway [Heuristic]:** the settlements reduce risk for *major-label* catalog exposure but do not immunize Suno output generally; training-data litigation and the human-authorship problem remain live. Advise clients that Suno output carries residual legal uncertainty and that high-stakes commercial uses (national ad, sync, brand campaign) warrant legal review and, where possible, human-authored components. --- ## 6. The API situation (verify before recommending any integration) **[Documented, verified 2026-07-10]:** - Suno has historically had **no official public API** — no developer console, no self-serve key page, no public SDK, no published docs. - On **2026-07-01**, Suno CPO Jack Brody announced Suno is **opening an API partner program**, starting with a **curated group of partners** (application via an intake form), framed as a precursor to a "partner-powered model." **No public launch timeline** was given, and it is **not** self-serve as of 2026-07-10. - **Third-party "Suno APIs" (e.g. sunoapi.org, PiAPI, gcui-art/suno-api and similar) are unofficial.** They work by reverse-engineering Suno's private web endpoints or by automating a logged-in browser (often with CAPTCHA-solving). They are **not authorized by Suno** and their use generally **violates Suno's Terms of Service** (which prohibits unauthorized automated access). - **[Heuristic / advisory]** "Commercial license" guarantees advertised by unofficial API resellers are **not something they can lawfully grant**, because they themselves operate outside Suno's sanctioned ecosystem, and the underlying copyright caveat (§5.2) still applies. If a production team needs programmatic Suno access today, the compliant path is to **apply to the official partner program** and wait, not to wire up a scraper. Flag any request to build on an unofficial Suno API as a ToS-violation and legal risk before proceeding. --- ## 7. Safety, consent, and rights hygiene - **Real-artist voices/likeness:** Only use an artist's voice, name, or likeness where they have **opted in** through the official WMG (or comparable) program, or where you otherwise hold rights. Generating "make it sound like [named living artist]" for release is a rights and publicity-rights risk even when the audio is AI-made. - **Lyrics you don't own:** Don't feed copyrighted lyrics (or melodies via Cover) you have no license to. Distributors will pull tracks and blacklist accounts for it. - **Disclosure:** Comply with DSP AI-disclosure (DDEX) at upload; don't pass AI tracks off as human-recorded where disclosure is required. - **Don't overstate ownership:** When advising, separate "you can use it commercially" (true on paid tier) from "you own an enforceable copyright" (often false for pure AI output). - **Not legal advice:** For a specific commercial deal, sync placement, or campaign, route the client to a qualified music/IP attorney. This skill informs; it does not replace counsel. --- ## 8. Worked examples (illustrative, not required formulas) ### Example A — Release-ready pop track for a paid client **Intent:** upbeat indie-pop single a small brand will use in a monetized YouTube ad. **Tier decision:** must generate on **Pro/Premier** (commercial rights require paid-at-creation). Write original lyrics in-house to add a human-authored copyright layer. **Style field:** `indie pop, bright and hopeful, 112 bpm, jangly electric guitar, warm female vocal, punchy live drums, radio-ready mix` **Exclude styles:** `lo-fi, trap, autotune-heavy` **Lyrics field (human-written):** ``` [Verse 1] <your original lyrics> [Pre-Chorus] <your original lyrics> [Chorus] <your original, repeated hook> [Verse 2] ... [Bridge] ... [Chorus] <same hook> [Outro] ``` **Why:** paid tier secures commercial use; original lyrics improve the human-authorship position; short structural tags give a clean radio arrangement. **Expected:** a clean, distributable track. **Failure modes:** genre words leaking into lyrics if placed in the wrong field; pure-AI melody may still be **non-copyrightable** (§5.2). **Rights note the agent must surface:** flag DDEX AI-disclosure at DistroKid upload; note Content ID is unavailable on the AI audio; note residual litigation uncertainty. ### Example B — Instrumental bed for a podcast (budget-conscious) **Intent:** 60–90s calm instrumental loop for a personal, non-monetized hobby podcast. **Tier decision:** if truly non-commercial and personal, free tier is *permissible* but requires **attribution to Suno** and forbids monetization; if the podcast has any sponsorship, it's commercial → **must be paid tier**. Clarify monetization before choosing. **Style field:** `ambient lo-fi, calm and reflective, 80 bpm, soft Rhodes piano, mellow tape hiss, gentle vinyl crackle, no drums` **Lyrics field:** `[Instrumental]` (or use the Instrumental toggle) **Why:** naming instruments and "no drums" plus `[Instrumental]` reliably yields a vocal-free bed. **Failure modes:** stray vocals if the instrumental flag isn't set; free-tier output can't be used the moment the show takes a sponsor. ### Example C — Client asks "wire Suno into our app via an API" **Correct response:** Explain there is **no official self-serve API** as of 2026-07-10; Suno opened a **curated partner program (announced 2026-07-01)** you can **apply** to. Warn that third-party "Suno APIs" are **unofficial scrapers that violate Suno's ToS** and cannot lawfully guarantee commercial licenses. Recommend applying to the official program and, meanwhile, not building on a scraper. **Critical failure:** recommending a third-party Suno API as if it were official/authorized, or promising it confers commercial rights. --- ## Sources (verified 2026-07-10) Official / first-party: - Suno Model Timeline — https://help.suno.com/en/articles/5782721 - Suno Terms of Service (revised 2026-03-26) — https://suno.com/terms-of-service - "Does Suno own the music I make?" — https://help.suno.com/en/articles/2416769 - "Do I have the copyrights to songs I made?" — https://help.suno.com/en/articles/2746945 - "Can I distribute my songs to Spotify, etc?" — https://help.suno.com/en/articles/2410177 - Suno v5.5 blog — https://suno.com/blog/v5-5 - Suno release notes — https://suno.com/release-notes Industry press (secondary, labeled): - Suno–WMG settlement/partnership — https://www.rollingstone.com/music/music-features/suno-warner-music-group-ai-music-settlement-lawsuit-1235472868/ - 2026 changes under Warner deal — https://www.digitalmusicnews.com/2025/12/22/suno-warner-music-deal-changes/ - Suno API partner program — https://www.musicbusinessworldwide.com/suno-explores-developer-api-seeking-apps-that-unlock-experiences-generative-music-makes-possible-for-the-first-time/ and https://www.digitalmusicnews.com/2026/07/03/suno-is-opening-an-api-partner-program/ - UMG–Udio settlement — https://www.rollingstone.com/music/music-features/ai-music-universal-music-group-settlement-udio-1235457945/ - UMG–Suno settlement impasse (Apr 2026) — https://www.digitalmusicnews.com/2026/04/09/suno-universal-music-lawsuit-settlement-impasse/ - AFM suit over settlements — https://www.hollywoodreporter.com/music/music-industry-news/musicians-union-lawsuit-ai-song-generator-settlement-1236614835/ - V5 launch coverage — https://musically.com/2025/09/25/suno-launches-v5-claiming-its-the-worlds-best-music-model/
© calesthio · MIT · calesthio/generative-media-skills
Похожие скиллы
Все скиллы по теме «ИИ и ML» →Сначала — suno-music от других авторов.