@kolbo/mcp 1.89.1 → 1.89.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kolbo/mcp",
-  "version": "1.89.1",
+  "version": "1.89.3",
   "description": "Kolbo AI MCP Server - Generate images, videos, music, speech, and sound effects from Claude Code",
   "main": "src/index.js",
   "bin": {

package/skill/GENERATED.md

@@ -1,6 +1,6 @@
 # AUTO-GENERATED — do not edit
 
-This tree is mirrored from kolbo-code@1dc58c1, the single source of truth.
+This tree is mirrored from kolbo-code@73b21c8, the single source of truth.
 Canonical source: packages/opencode/skills/kolbo/
 Distribution: .github/workflows/sync-skill-to-plugin.yml
 

package/skill/references/models/music.md

@@ -13,7 +13,13 @@ Load this file when the user wants AI-generated **music** — full songs, lyrics
 ## CRITICAL Kolbo Platform Rules
 
 - **Model version, duration, and instrumental toggle are MCP-tool params.** Don't write `v4.5`, `30 seconds`, or `instrumental: true` inside the prompt fields themselves.
-- **Exact track length = `duration_seconds`** (clamped 5–300s). Only length-controllable models honor it (e.g. ElevenLabs Music, `music-v1`) — without it those models default to a ~10s track, so ALWAYS pass it for jingles/beds on those models. Suno ignores it and picks its own length.
+- **Suno v6 is the current family**: `suno-v6` (the default music model), `suno-v6-mini`, `suno-v6-wild` — all **15 credits**, identical to the older Suno rows. `suno-v5.5` / `suno-v5` / `suno-v4.5plus` are still selectable, but the provider marks V4 / V4_5 / V4_5PLUS / V4_5ALL / V5 / V5_5 as discontinued upstream — a dead branch. Pick v6 unless the user asks for an older take.
+- **Exact track length = `duration_seconds`** (clamped 5–300s on the tool). **Suno is length-controllable as of v6** — 10 seconds to 6 minutes, slider or an exact typed `m:ss`. **Auto is the default**, and on Auto Suno picks its own natural length exactly as it always did, so only pass a length when the user actually asked for one.
+  - **Custom Mode only.** The provider returns a 422 if a duration arrives in Simple Mode, so length only lands alongside lyrics — the SDK/MCP path switches the request into Custom Mode automatically when `duration_seconds` is present.
+  - **`suno-v6-mini` does NOT honour length.** Measured, not assumed: 20s / 30s / 60s requests all came back ~200–220s. Never promise length control on Mini. `suno-v6` and `suno-v6-wild` honour it.
+  - Non-Suno length-controllable models (ElevenLabs Music, `music-v1`) still default to a ~10s track without it, so ALWAYS pass it there for jingles/beds.
+- **Short lengths are style-dependent — write a sparse prompt when you need a short clip.** Suno resolves a musical phrase before it stops: a sparse ambient prompt asked for 13s returns ~13s, a dense rock prompt asked for the same 13s returns ~42s. They converge by about 30s. For a genuinely short cue, thin the instrumentation in the `style` field instead of fighting the number.
+- **Suno usually returns TWO tracks, and does not always apply the requested length to both** — one can land on target while the other runs long. Check each track before delivering; the first track's length is not authoritative.
 - Suno generations have **two separate input fields**: a **Style / Description** field (`style` param) and a **Lyrics** field (`lyrics` param). Output your prompt as **TWO separate fenced code blocks** so the user (and the tool call) know exactly what goes where.
 - Tell the user to run the prompt multiple times — Suno output varies significantly between generations, that's a feature. Use `num_generations` if the tool supports it, or fire 2–4 parallel `generate_music` calls.
 
@@ -88,7 +94,7 @@ Use Suno's section tags to control structure. Each tag goes on its own line, con
 ### Jingle / ad music (15–30s)
 - `style`: short, punchy descriptor (`upbeat retail pop jingle, female vocal, claps, glossy production, summer energy`)
 - `lyrics`: 2–4 short lines max, often just chorus
-- Pass the shortest `duration` the tool supports — or, on a length-controllable model (ElevenLabs Music), pass the exact `duration_seconds` (e.g. `15` or `30`).
+- Pass the exact `duration_seconds` (e.g. `15` or `30`). Suno v6 / v6 Wild honour it in Custom Mode (supply `lyrics`), ElevenLabs Music honours it either way, `suno-v6-mini` does not. Keep the style sparse — a dense arrangement overruns a short target.
 
 ### Cinematic trailer / score
 - `style`: `cinematic orchestral trailer, swelling strings, taiko drums, hybrid choir, dramatic build, modern hybrid score`
@@ -116,7 +122,7 @@ LYRICS:
 
 When summarizing to the user, state separately:
 - **Instrumental:** yes / no (the `instrumental` param)
-- **Recommended duration:** short / medium / long (the `duration` param)
+- **Length:** Auto, or the exact `duration_seconds` you passed (10s–6min; Suno needs Custom Mode, and `suno-v6-mini` ignores it)
 - **Run takes:** N generations (usually 2–4) — fire them in parallel
 - **Why this works:** 1 line on the key genre / structure / instrumentation choice
 

package/skill/references/models/seedance.md

@@ -36,6 +36,8 @@ Load this file when the user wants a **Seedance 2 / Seedance 2.0** (ByteDance) v
 
 ## Locked Intro (DEFAULT for any multi-shot cinematic — including Elements)
 
+**Write every Seedance prompt in TWO LAYERS.** Layer 1 is the locked blocks below: everything the scene holds CONSTANT — shooting style, grade, fixed lighting, fixed props/elements, cast (look AND character), location, the location's physical scale, the piece's movement speed and assertiveness, and where each performer stands relative to the location and to each other. Layer 2 is the timecoded SHOT list: only what CHANGES, in full detail, shot by shot. A constant restated per shot wastes the character budget; a change buried in Layer 1 never happens.
+
 After the Total lines, every multi-shot prompt — and any piece with recurring people or a recurring place — opens with the locked blocks. Skip only for: true single-shot POV/orb, 3×3 grid-panel mode, or video-edit tasks.
 
 ```
@@ -46,11 +48,11 @@ Total: Xs / N shots / AR
 <body>, <lens family>, <film stock>, <aspect> spherical, <stop>. <DoF, grain, grade as law>. <movement grammar>. <performance + audio law>.
 
 [CAST – IDENTICAL IN EVERY SHOT]
-@Exact_DNA_name: age, build, hair, face, wardrobe, signature details.
+@Exact_DNA_name: age, build, hair, face, wardrobe, signature details. PERSONA: emotional baseline, energy, posture/gait, habitual gestures, how much they exaggerate.
 PROP: recurring object.
 
 [LOCATION]
-Place in materials + light + color field. Blocking. Background LIFE.
+Place in materials + light + color field. Physical SCALE (room/street size, ceiling height, distances to cover). Blocking — who stands where, and on which side of whom. Background LIFE.
 
 [LOCATION MAP]
 Named seats / sides in SCREEN language (screen-left armchair, center couch, door camera-left).
@@ -125,6 +127,14 @@ These elevate rich cinematic / reference-anchored sequences. For a short, tight,
 - **AVOID** — add when there's real drift risk: referenced characters/DNAs, an established story state, or a specific content tone to enforce. `AVOID:` + the drifts to forbid: character morphing, costume/anatomy changes, missing signature details (a piercing, a scar), camera looking into the lens, glamorous posing, duplicated characters, extra weapons, text / subtitles / logos, plus content-specific negatives (e.g. "intact gemstone, red aura" once it's destroyed). Skip it when there's nothing meaningful to forbid.
 - **PHYSICAL GROUNDING** — only when the shot has a real contact/impact (a boot planted on a chest, a weapon lock, a body thrown): spell out weight transfer, full contact, stable balance, material reaction, and the victim's physical response. Not needed for talking, walking, or ambient shots.
 
+## Persona & performance (HARD — any living character: people, animals, creatures)
+
+Appearance locks WHO. Persona locks HOW THEY BEHAVE — without it Seedance renders a technically correct robot.
+
+- **Layer 1:** every living cast member gets a PERSONA clause in CAST — emotional baseline (warm / guarded / furious under control / anxious), energy level, posture and gait, habitual hand gestures, exaggeration level. Lock it once and reuse the SAME persona text verbatim on every later generation in the project, so the character does not change personality between clips.
+- **Layer 2:** every shot beat with a living character states the PERFORMANCE, not only the movement — facial expression, the emotion under it, and how that emotion shapes the physical action. A happy walk, a frightened run and an angry line reading are three different bodies. Mixed states are usually truer ("relief fighting embarrassment"). Add hand gestures sized to the locked persona.
+- Forbidden: a beat that moves a person through the frame with no expression, no emotion and no energy. That is the single most common cause of "the video looks dead".
+
 ## Dialogue & expression
 
 - **Dialogue is PERFORMED by the model, never by a TTS tool.** Quoted lines in the prompt come back as synced speech with lip movement and room tone, together with the SFX you name in AUDIO. Scene dialogue therefore never routes through `generate_speech` or `generate_lipsync` — write the line in quotes inside its shot beat and let Seedance act it.

package/skill/references/models/seedance25.md

@@ -38,6 +38,10 @@ Load this file when the user wants a **Seedance 2.5** video (they said "2.5" / "
 - Duration range **4–30s**; shot count **≤30** in one generation. Do not split a ≤30s story into multiple 10s clips unless the user asks.
 - Omit Total / Multishot / shot-count headers only for **video editing** (source duration locked) — use Edit Goal blocks instead.
 
+## Two layers (HARD)
+
+Layer 1 = the Locked Intro: everything constant — shooting style, grade, fixed lighting, fixed elements/props, cast look AND persona, location, the location's physical scale, the piece's speed and assertiveness, and each performer's position relative to the location and to the other performers. Layer 2 = the timecoded SHOT list: only what changes, in full detail. Persona and per-shot performance rules: `models/seedance.md` § Persona & performance — at 30s and 30 shots a cast with no locked persona drifts into a different person by the last cut.
+
 ## Locked Intro (DEFAULT — same shape as Seedance 2)
 
 ```