# Suno > This skill should be used when the user wants to compose songs for Suno AI, write lyrics, create style prompts, or generate Suno metatags. Supports J-pop, K-pop, EDM, ballads, rock, and Latin genres, plus album/EP composition, variations, and song continuations. Also handles reference-based composition ("like YOASOBI", "in the style of Aimer") and J-pop tier presets ("anisong", "mainstream", "doujin"). Triggers on "write a song", "Suno prompt", "Suno metatags", "style of music", "song lyrics", "Suno AI", "acoustic version", "create an album", "extend this song", "like [artist]", "in the style of", "/suno", "anisong", "doujin", "negative prompting", "ad-libs". Use when this capability is needed. - Skill: `tomevault-io/suno` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/suno` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/suno/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/suno --- # Song Composition for Suno AI Targets current Suno models (v5/v5.5). Where v5.5 behavior is unverified, guidance says so and hedges rather than guessing. ## Creative Engine Role Act as a **songwriter first**, not a rule-follower. The references in this skill are a creative palette — inspiration, not prescription. | Reference | Use it as... | NOT as... | |-----------|--------------|-----------| | Artist profiles | Vibe inspiration | Exact tag lookup | | Genre conventions | Starting points | Rigid rules | | Metatags | Toolkit of options | Required checklist | | Tier presets | Creative directions | Auto-apply templates | **Creative latitude:** Blend genres unexpectedly. Interpret "like [artist]" through artistic essence, not exact specs. Break conventions when the song calls for it. The Composition Rubric below is the **floor** (craft that must hold), never the ceiling (taste stays yours). ## Style Prompts — Dual Form Suno v5.5 guidance favors modular descriptor lists; this plugin's narrative form was first-party-tested on v5 and is unverified on v5.5. Until a live A/B settles it, **output BOTH forms** for every song (the `style-prompt form` preference can pin one): **Modular — label it "try first":** 4-7 weighted descriptors in this order — genre+subgenre · mood/energy · vocal direction · key instruments · production/tempo. One clear genre anchor; no near-synonym stacking. ``` gothic symphonic metal, doujin style, dramatic and tragic, dual female vocals in dialogue with soaring harmonies, harpsichord pipe organ strings and double-kick drums, polished cathedral-reverb production at 172 bpm ``` **Narrative — label it "fallback if the arrangement feels flat":** prose describing how the song unfolds in time ("opens with… enters… builds through… resolves into"). Its v5-tested strength is temporal arrangement control. Weave the emotion arc into the narrative; don't append it. Rules that apply to **both** forms: - **Top-anchor:** lead with the vocal persona (…warm female vocals with melismatic ornamentation…) — UNLESS a Suno Voice/Persona/Custom Model is attached, in which case **omit vocal descriptors entirely** (they conflict with the attached voice). - **Single genre anchor** — one clear genre reference, not a pile of cousins. - **Exclusions go in Suno's Exclude (Styles) field**, not the style text: ≤2 entries, each paired with a replacement ("no lead guitar solo — fingerpicked acoustic instead"). See `references/suno-metatags.md`. Arrangement precision lives in the lyrics field via **parameterized section tags** (next section), not in the style prompt. ## Metatags in Lyrics Suno reads embedded directions in the lyrics field. **Tag sparsely: 3-4 inflection points per song.** Most sections get only the bare marker — verse/chorus structure already creates contrast. Current syntax is **parameterized section tags**: `[Bridge: stripped, harpsichord and voice]`, `[Final Chorus: key change up]`. The adjacent-bracket form (`[Bridge][stripped]`) is legacy-valid. | Moment | Purpose | Example | |--------|---------|---------| | **Intro** | Set opening texture | `[Intro: music box and choir, then full band]` | | **Breakdown/Bridge** | Contrast point | `[Bridge: stripped, half-time]` | | **Build** | Pre-climax tension | `[Build]`, `[Build: snare roll]` | | **Final Chorus** | Earned peak | `[Final Chorus: key change up]` | Use **technique cues** (`stripped`, `half-time`, `key change up`) not **emotion words** (`triumphant`, `intimate`). Vocal technique tags (`whisper`, `belting`, `falsetto`) are the exception — they describe HOW to sing. A `[Break]` or blank line before the climax buys tension through silence. **Paste-clean contract:** the Lyrics block contains ONLY what Suno should sing or obey — no kana counts, no voice arrows, no notes-to-self. Parentheticals in lyrics are SUNG as ad-libs. All craft metadata (voice map, reading declarations, mora counts) lives in the Readings & Casting block of the output format (`references/output-formats.md`). ## Lyric Writing **English:** target 6-10 syllables per line; line breaks = musical breaths. **Japanese:** count **morae, not syllables** (月光 = げっこう = 4 morae). Hook lines 10-14 morae, verse lines 9-14, parallel-couplet hemistichs 3-8; held notes land on open or long vowels; register devices for literary subjects. Full rules: `references/japanese-prosody.md` — consult it before writing any Japanese lyric. **Mixed language:** switch at phrase boundaries; English hooks suit mainstream/idol/anisong registers, not doujin/literary ones. See `references/suno-metatags.md` for ad-libs, punctuation cues, and vowel elongation. ## Composition Rubric (the floor — lyric-critic enforces this) Checkability tiers: **[M]echanical / [S]tructural** violations are findings; **[J]udgment** concerns are advisory only — taste is not agent-verifiable. 1. **[S] Voice casting** — duet/dual vocal config ⇒ the lyric assigns voices (dialogue verses, unison/harmony chorus, descant or trade-off finale), recorded in Readings & Casting. 2. **[M] Prosody** — line counts within the targets above, counted against DECLARED readings; section-peak held notes on open/long vowels. 3. **[S] Register** — literary/gothic/historical subject ⇒ ≥2 register devices. Japanese devices: `references/japanese-prosody.md`. English: archaic pronouns, latinate diction, syntactic inversion. 4. **[S] Imagery system** — one concrete image system from the subject's lexicon; stock-word cap per register (strict for doujin/literary/gothic, advisory for mainstream/idol/anisong; pronouns exempt — see japanese-prosody.md). 5. **[S] Thesis presence** — the chorus contains one candidate line carrying the song's emotional argument. **[J]** Whether it lands concretely is advisory. 6. **[M] Hook anchor** — title or hook phrase opens or closes the chorus. 7. **[M] Tags** — 3-4 parameterized inflection points, technique cues only. 8. **[M] Style prompt** — both forms present (or preference-pinned single form); vocal descriptors absent when a Voice is attached; ≤2 Exclude entries with replacements; Lyrics block paste-clean. Quality bar in action: `references/examples/gothic-doujin-song.md` (every rubric item annotated). ## Genre Conventions > Springboards, not rules. Blend and break as the song demands. | Genre | Key Elements | Common Tags | |-------|--------------|-------------| | J-pop | Catchy hooks, key changes, electronic+acoustic | j-pop, catchy melody, anime | | Doujin | Electronic, 140-180 BPM, dramatic shifts | vocaloid style, fast tempo | | Ballad | 60-90 BPM, piano/guitar lead, emotional | ballad, piano, heartfelt | | Rock | Guitar-driven, powerful vocals | j-rock, electric guitar | | EDM | Build-drop patterns, synth-led | edm, electronic, dance | | K-pop | Polished, genre-blending | k-pop, polished, hook-driven | | Latin | Distinctive rhythms, Spanish/Portuguese | latin, reggaeton, tropical | For subgenre depth and the romanized phrase banks, see `references/genre-deep-dive.md`. Song structure templates: `references/song-structures.md`. ## Mood-to-Style Mapping | Mood | Tempo | Tags | |------|-------|------| | Upbeat | 120-140 | energetic, bright, cheerful | | Melancholic | 70-90 | sad, emotional, bittersweet | | Energetic | 140-170 | driving, intense, anthemic | | Dreamy | 80-100 | atmospheric, ethereal, floating | | Intense | 130-160 | dramatic, dark, cinematic | | Chill | 85-110 | relaxed, smooth, laid-back | ## Tempo Guidelines | Style | BPM Range | Feel | |-------|-----------|------| | Slow ballad | 60-75 | Intimate, emotional | | Mid-tempo ballad | 75-95 | Flowing, expressive | | Pop standard | 100-120 | Comfortable, catchy | | Dance pop | 120-135 | Energetic, groovy | | Fast pop/rock | 135-160 | Driving, exciting | | High energy | 160-180 | Intense, powerful | Vocal tag vocabulary (female/male/special/tone/effects): `references/suno-metatags.md`. ## Reference-Based Composition When a user says "like YOASOBI" or "in the style of Aimer", capture the artist's *essence* — creative spirit, emotional signature, sonic identity — not exact specifications. Ask: what FEELING does this artist evoke? What makes them recognizable? For technical grounding, consult `references/artist-profiles.md` (profiles tagged by tier, including `Register:` lines for lyric-register matching). The profile gives ingredients; you decide the recipe. **If the profile lists a dual/duet vocal configuration, write FOR it** — rubric item 1. For unknown artists: interpret from what you know and capture the requested spirit. ## J-pop Tier Presets Ecosystem-level presets instead of specific artists: | Tier | Keywords | Sound | |------|----------|-------| | **Anisong** | `anisong`, `anime` | Anime OP/ED - dramatic, catchy, high energy | | **Surface** | `surface`, `viral` | Producer scene - complex, narrative, layered | | **Mainstream** | `mainstream`, `normie` | Radio-friendly - accessible, sing-along | | **Doujin** | `doujin`, `touhou` | Convention - high production, niche genres | | **Legacy** | `legacy`, `city pop` | Golden age - warm analog sound | Combine with artists: `/suno anisong like Aimer` uses tier structure + artist style. Tier presets are **ingredient lists** — render them through the dual-form style prompt rules above. Full profiles and merge logic: `references/jpop-tiers.md`. ## Professional Songwriter Techniques **Hook-First:** design the chorus hook before the verses that build to it. **Tension & Release:** Verse (low) → Pre-Chorus (building) → Chorus (peak) → Post-Chorus (release). **Three-Element Arrangement:** A (melody), B (counter-melody), C (rhythm). Verse: A+C. Chorus: A+B+C. Bridge: B+C. Details and exercises: `references/pro-techniques.md`. ## Output Formats All output templates — including the **Readings & Casting craft block** and the dual-form style prompt slots — live in `references/output-formats.md`. ## Additional Resources - **`references/output-formats.md`** - Song/album/variation templates, craft block - **`references/suno-metatags.md`** - Metatags, parameterized tags, Exclude field, formatting - **`references/japanese-prosody.md`** - Mora rules, register devices, stock-word list - **`references/song-structures.md`** - Standard pop, anime, ballad structures - **`references/pro-techniques.md`** - Hook-first, tension/release, arrangement - **`references/genre-deep-dive.md`** - Subgenre conventions and phrase banks - **`references/artist-profiles.md`** - Artist profiles for reference-based composition - **`references/jpop-tiers.md`** - Anisong, surface, mainstream, doujin tiers - **`references/album-composition.md`** - Album arc patterns and track roles - **`references/variation-patterns.md`** - Transformation matrices for variations - **`references/continuation-patterns.md`** - Callback techniques for extensions - **`references/examples/gothic-doujin-song.md`** - Annotated quality-bar exemplar - **`references/troubleshooting.md`** - Common issues, causes, and fixes --- > Source: [Linaqruf/kana-code-plugins](https://github.com/Linaqruf/kana-code-plugins) — distributed by [TomeVault](https://tomevault.io).