@kolbo/mcp 1.70.0 → 1.70.2

package/skill/references/filmmaking/routing.md

@@ -0,0 +1,105 @@
+# Routing
+
+Use this file to choose the minimum effective workflow and context.
+
+## 1. Select the production stage
+
+| Stage | User intent | Primary deliverable |
+|---|---|---|
+| Development | Invent, structure, write, or repair story material | Premise, outline, scene, sequence, screenplay pages, structural audit |
+| Pre-production | Prepare stable inputs before spending generations | Production bible, asset plan, character/location/prop/voice/state specification, storyboard |
+| Direction | Turn an approved shot or scene into executable model instructions | Shot card, generation prompt, blocking map prompt, audio plan |
+| Audit | Diagnose a prompt or generated result | Root cause, failed contract, smallest corrective action |
+| Workbench | Revise a known prompt after a failed take | Section-level patch plus unchanged remainder |
+| Production | Coordinate many shots and downstream editorial needs | Shot list, continuity ledger, generation log, coverage/QC status |
+
+If the user asks for several stages, preserve order: Development → Pre-production → Direction → Audit/Workbench → Production/QC.
+
+## 2. Select the generation family and mode
+
+Decide from inputs and intended output, not from the model brand.
+
+| Intent and inputs | Mode |
+|---|---|
+| Existing source video must change | video-to-video |
+| Required start and end images | first/last frame |
+| One still is the exact opening frame and must animate | image-to-video |
+| Character, location, product, prop, diagram, or multimodal references define a new scene | elements/reference-driven |
+| Explicit words-only generation, no references | text-to-video |
+| Source waveform drives performance | audio-driven/lip-sync capable mode |
+
+Start/end-frame intent outranks Elements. Existing-video edit outranks fresh generation. Do not ask which mode when the evidence is sufficient.
+
+## 3. Select control density
+
+### Strict
+
+Use when one wrong detail invalidates the take:
+
+- dialogue and lip ownership;
+- precise multi-person blocking or 180-degree axis;
+- prop hand, count, scale, state, injury, or contact;
+- exact timing, choreography, lyrics, or cut points;
+- repeated generation failure;
+- expensive hero shot or cross-shot continuity seam.
+
+Specify first frame, time windows, positions, camera side, physical causality, dialogue ownership, and explicit failure locks. Keep each time window light enough to execute.
+
+### Anchored
+
+Use when key facts must hold but variation may improve the result:
+
+- spectacular action with known start/end anchors;
+- creature or vehicle motion where physics and scale must hold;
+- reveal, rescue, transformation, or impossible shot;
+- cinematic coverage where exact camera path is not essential.
+
+Lock identity, geography, state, essential beats, physics, and final state. Grant freedom only to named dimensions such as camera path, shot size, or moment-to-moment staging.
+
+### Exploratory
+
+Use when variation is the point:
+
+- montage and b-roll;
+- music-video inserts and alternate coverage;
+- mood exploration;
+- early visual development.
+
+Keep identity, world, period, audio ownership, safety, and editorial objective fixed. Invite a bounded range of camera or performance solutions.
+
+## 4. Select the audio lane
+
+| Lane | Use when | Required truth |
+|---|---|---|
+| Dialogue | Actors speak exact lines | speaker, verbatim text, timing, voice identity, listener behavior, silence ownership |
+| Exact-song lip-sync | A supplied waveform must own the mouth | source asset, performer ownership, exact lyrics/phonetics, breath seams, non-performer mouth behavior |
+| Native music performance | The model should create or perform a musical moment | musical structure, performers, vocal ownership, tempo/beat behavior, desired score/SFX relationship |
+| Ambience/SFX only | Music belongs in post or silence is dramatic | environment bed, timed effects, no-score instruction if model supports it |
+| Post-production music | Picture should remain edit-friendly | continuous ambience and clean dialogue/SFX; score brief tracked separately |
+
+Do not turn “no music” into “no audio.” Preserve requested dialogue, ambience, and effects.
+
+## 5. Select craft packs
+
+Load a pack only when a relevant signal is present:
+
+- **Story:** scene, sequence, goal, stakes, reversal, value shift, screenplay, weak dramatic beat.
+- **Assets:** recurring identity, new state, location plate, prop, vehicle, crowd, character sheet, stress test.
+- **Acting:** speaking, listening, reaction, close-up, dead eyes, flat performance, overacting, subtext.
+- **Blocking:** more than one subject, dialogue coverage, recurring location, screen direction, position drift, staging map.
+- **Cinematography:** shot size, lens, camera path, style, reveal, edit grammar, perspective.
+- **Physics:** contact, weight, vehicle, creature, water, particles, weapon, fall, jump, transformation, impossible gravity.
+- **Audio:** speech, song, lip-sync, score, sound design, ambience, voice continuity.
+
+## 6. Select the adapter
+
+Read the target model adapter last so it can translate the creative plan into actual limits and syntax. Never let the adapter replace the story, performance, or production truth.
+
+If the model is unspecified, ask only when the answer materially changes the deliverable. Otherwise produce a model-neutral shot card and state which adapter remains unresolved.
+
+## 7. Select output depth
+
+- If the user asks only for a prompt, lead with the prompt and keep notes short.
+- If the user asks for an audit, give the root cause and patch, not a complete rewrite unless required.
+- If the user asks for a workflow, produce saved-state artifacts before generation prompts.
+- If essential production truth is missing and guessing would risk paid generations, ask at most three targeted questions. Otherwise choose defensible defaults and label them outside the prompt.

package/skill/references/filmmaking/scene-engine.md

@@ -0,0 +1,95 @@
+# Scene Engine
+
+Use this causal engine for writing and auditing dramatic scenes and sequences.
+
+## The chain
+
+```text
+Goal → Obstacle → Tactic → Reversal → Audience Value Shift
+```
+
+Apply each term precisely.
+
+### Goal
+
+Define the single result the protagonist is fighting toward. Keep the major goal stable. Make each scene goal a necessary causal link toward it.
+
+Use the removal test: if deleting the scene leaves the causal path intact, the scene lacks a necessary goal or belongs elsewhere.
+
+Do not confuse the goal with what the hero “ought” to do. The obstacle and stakes force the next tactic.
+
+### Obstacle
+
+Define a circumstance that jeopardizes a stage of the path or the whole goal. Name what is at risk and whether the scale is local or global.
+
+A cost is not automatically an obstacle. The obstacle must threaten failure strongly enough to force a choice or tactic.
+
+### Tactic
+
+Define a plausible move based on the character's current incomplete knowledge. Treat failure as search under uncertainty, not stupidity.
+
+Every failed tactic must return information, narrow the search, alter power, or reshape the next move. Repetition that teaches nothing is wheel-spinning.
+
+### Reversal
+
+Create a turn against expectation:
+
+- the character's own action flips against its purpose;
+- hidden agency is revealed;
+- the situation turns opposite to what the tactic intended.
+
+Require at least one meaningful reversal in each resolved sequence. A sequence begins when a particular jeopardy opens and ends when it is overcome or defeats the character and forces a new path. Sequences may nest.
+
+### Audience value shift
+
+Name how the reversal changes the audience's verdict on a character:
+
+```text
+before verdict → after verdict
+```
+
+If a plot turn produces no changed judgment, treat it as inert. Design value shifts to deepen a portrait rather than randomly oscillate.
+
+## Scene development
+
+Before writing, establish:
+
+1. Story goal and the scene's causal link.
+2. Open jeopardy and scale.
+3. Current knowledge/truth gap.
+4. Tactics and information returned by each.
+5. Sequence reversal and form.
+6. Audience verdict before and after.
+7. The final beat that makes the event legible.
+
+For screenplay pages, write clean present-tense dramatic action and standard screenplay formatting. Let behavior carry subtext.
+
+## Structural audit
+
+Report:
+
+```text
+SCENE / SEQUENCE
+Goal — causal link and removal test
+Obstacle — what is jeopardized and at what scale
+Tactic — why it is plausible and what it returns
+Reversal — form and placement
+Value shift — audience verdict before → after
+Weakest point — the single highest-leverage failure
+Minimal repair — preserve the user's design
+Clean repair — rebuild enough to make the chain work
+```
+
+Do not demand a reversal in every individual shot. Audit it at the sequence scale while allowing individual scenes or beats to escalate pressure.
+
+## Bridge into directing
+
+Translate the structural engine into filmable direction:
+
+- scene goal becomes the dramatic job of the coverage;
+- obstacle becomes visible pressure in space, time, behavior, or consequences;
+- tactic becomes playable action toward another person or object;
+- reversal gets a visual or sonic event and an editorial emphasis;
+- value shift gets the reaction image that changes the audience's read.
+
+Never paste structural analysis into a generation prompt unless it produces observable action, performance, blocking, camera, or sound.

package/skill/references/filmmaking/validation.md

@@ -0,0 +1,109 @@
+# Validation and Quality Control
+
+## Pre-generation audit
+
+### Story and edit
+
+- Does the shot have a necessary dramatic/editorial job?
+- Is the intended reaction, reveal, reversal, or information change visible/audible?
+- Does the start/end state support the neighboring cuts?
+
+### Assets and references
+
+- Does every tag exist in the registry?
+- Is each exact state variant correct?
+- Are inactive/stale tags removed?
+- Does each reference have a declared role?
+- Are input/reference limits respected?
+- Should a difficult condition become a new asset instead of prose?
+
+### Geography and continuity
+
+- Are first-frame positions, facing, eyelines, camera side, and prop hands clear?
+- Can every action and relocation happen in time?
+- Do internal cuts preserve or deliberately change state?
+- Is final state explicit for the next shot?
+
+### Acting
+
+- Does each important character pursue a playable tactic?
+- Do listeners have tasks?
+- Are beat changes motivated and observable?
+- Is performance direction behavior-first rather than emotion-label-first?
+- Is voice identity stable?
+
+### Camera and light
+
+- Does lens/FOV match content?
+- Is operator movement physically coherent?
+- Do camera freedom and strict blocking conflict?
+- Is light source/direction/exposure coherent across cuts?
+- Does style support rather than overwrite the scene?
+
+### Physics
+
+- Are mass, contact, support, force, clearance, and consequence plausible?
+- Are props/creatures/vehicles in one consistent state per beat?
+- Is complex action split or input-anchored when necessary?
+
+### Audio
+
+- Does one owner control each spoken/sung line?
+- Do exact words/lyrics and timing fit?
+- Is source/native/post audio policy clear?
+- Does “no music” preserve dialogue/SFX/ambience?
+- Is there an editorial audio seam?
+
+### Model contract
+
+- Do duration, shot count, prompt length, aspect/mode, and reference limits fit?
+- Does the output format/card tag match the selected mode?
+- Are continuous-take and cut instructions consistent?
+
+## Post-generation QC
+
+Review the entire output, not only a representative frame:
+
+- identity and state stability through time;
+- facial/hand integrity and performance timing;
+- mouth ownership and audio synchronization;
+- blocking, axis, screen direction, eyelines;
+- prop counts, hands, contact, and continuity;
+- physics, scale, creature/vehicle motion;
+- camera, focus, lighting, edge artifacts, flicker, warping;
+- dialogue accuracy, voice drift, invented audio;
+- editorial usability of first/last frames and audio tails.
+
+Record the exact timestamp and observable defect.
+
+## Failure ownership
+
+Classify one primary owner:
+
+- asset/reference;
+- wrong state or missing bible truth;
+- dramatic/shot design;
+- blocking/geography;
+- acting/task;
+- optics/camera;
+- timing/action density;
+- physics/material;
+- lighting/style;
+- audio/voice/music;
+- model capability/mode;
+- prompt contradiction/format;
+- stochastic variation after all contracts are sound.
+
+## Surgical iteration
+
+Change one causal variable when possible. State the expected proof before regenerating. Preserve the exact version that worked.
+
+Set an attempt threshold appropriate to cost and complexity. When the same failure repeats beyond it, escalate from wording to production redesign.
+
+## Severity
+
+- **Blocker:** unusable take—identity, safety, missing actor/action/dialogue, broken physics, wrong state, impossible continuity, capability violation.
+- **Major:** story/performance/edit meaning changes or visible slop likely noticed by viewers.
+- **Minor:** polish issue that can be retouched or accepted without changing meaning.
+
+Do not approve a feature-critical take with unresolved blockers. Distinguish static prompt confidence from real generated-media proof.

package/skill/references/filmmaking/workflows.md

@@ -0,0 +1,80 @@
+# Workflows
+
+## Single shot
+
+1. Establish editorial job and dramatic event.
+2. Resolve active assets/states and model mode.
+3. Choose control density and craft packs.
+4. Write a shot card.
+5. Compile the prompt.
+6. Audit before generation.
+7. Log result and update continuity only after approval.
+
+## Connected dialogue scene
+
+1. Audit the scene engine and ending.
+2. Name the shared event/direction.
+3. Build each character's motive, goal, obstacle, tactic, physical channel, and voice lock.
+4. Build one location map and coverage axis.
+5. Plan masters, singles, two-shots, reactions, inserts, and audio seams according to the edit—not by formula.
+6. Write separate shot cards with entering/leaving performance and prop states.
+7. Compile dialogue ownership and listener behavior into each shot.
+8. Assemble early and order missing coverage.
+
+## Music performance
+
+1. Choose exact-source lip-sync, native music performance, or post-music lane.
+2. Lock track ownership, section boundaries, lyrics, beat landmarks, formation, and performer mouths.
+3. Divide long music at breaths/musical seams when separate generations are needed.
+4. Choose strict synchronization for hero performance and exploratory coverage for inserts/b-roll.
+5. Preserve character identity and wardrobe while allowing differentiated personality inside synchronized choreography.
+6. Log whether source or generated audio is authoritative in the edit.
+
+## Impossible shot
+
+1. Define the signature moment and visible success proof.
+2. Simulate geometry, gravity, scale, action, and duration.
+3. Decide what should be baked into inputs: state asset, inversion, first/last frame, staging map, depth map, motion reference.
+4. Lock start state, essential physics, beat order, and final state.
+5. Grant bounded camera freedom if variation can help.
+6. Generate coverage candidates and edit the best physically consistent pieces when one perfect take is unnecessary.
+
+## Connected sequence
+
+1. Define sequence jeopardy, reversal, and value shift.
+2. Break into scenes/shots by causal and editorial necessity.
+3. Build a scene map, continuity entry/exit state, and audio bed.
+4. Assign each shot an editorial function and complexity rating.
+5. Compile prompts from shot cards.
+6. Assemble continuously; use the edit to request coverage and simplify failures.
+
+## Commercial or music video
+
+1. Define hook, product/artist identity, audience, platform/aspect, music ownership, and signature moments.
+2. Separate hero shots from exploratory coverage.
+3. Lock brand/product facts and text rendering to appropriate assets or post.
+4. Use controlled variety in camera, formation, and locations without drifting identity or product state.
+5. Plan deliverable cuts and safe crop zones.
+
+## Feature or episode
+
+1. Lock script/structure enough to expose asset and continuity needs.
+2. Build production bible, visual worlds, character/voice profiles, asset registry, and state list.
+3. Stress-test core cast in real locations/actions.
+4. Break script into scene cards and preliminary shot cards carrying direction and edit intent.
+5. Establish location coverage/maps and scene audio beds.
+6. Generate in scene blocks with version logs and surgical iteration.
+7. Edit in parallel; order coverage while regeneration is cheap.
+8. Run generation supervision before fine cut.
+9. Lock picture before cleanup, color unification, sound design, and score finishing.
+
+Do not attempt to generate a feature automatically from a paragraph. Automation should create reviewable stages and preserve human creative decisions.
+
+## Audit and Workbench
+
+1. Compare prompt, inputs, result, and intended proof.
+2. Classify the failure owner.
+3. Patch one section/variable.
+4. Preserve proven prompt text and settings.
+5. Re-run and log.
+6. After the configured attempt threshold, redesign the shot or asset rather than continuing prose iteration.

package/skill/references/models/gpt-image.md

@@ -6,7 +6,7 @@
 
 Load this file when the user wants a **GPT Image 2 / gpt-image-2** image (OpenAI). For other image models see `models/nano-banana.md`, `models/creative-director.md`, or `models/prompt-copilot.md`.
 
-**Kolbo MCP routing:** call `generate_image` (text-to-image) or `generate_image_edit` (edits with `source_images`). Pass `model: "gpt-image-2"` when the user named it. For photoreal photo edits that need identity lock or readable on-image text, default to `gpt-image-2` via `generate_image_edit` — consult `list_models({ type: "image_editing" })`. Do not auto-pick Flux.
+**Kolbo MCP routing:** call `generate_image` (text-to-image) or `generate_image_edit` (edits with `source_images`). Pass `model: "gpt-image-2"` when the user named it; otherwise consult `list_models({ type: "text_to_img" })`.
 
 ## CRITICAL Kolbo Platform Rules
 

package/skill/references/models/nano-banana.md

@@ -6,7 +6,7 @@
 
 Load this file when the user wants a **Nano Banana 2 (Gemini 3.1 Flash Image)** or **Nano Banana Pro (Gemini 3 Pro Image)** image. For other image models see `models/gpt-image.md`, `models/creative-director.md`, or `models/prompt-copilot.md`.
 
-**Kolbo MCP routing:** call `generate_image` or `generate_image_edit`. Pass `model: "nano-banana-2"` or `model: "nano-banana-pro"` when the user named one. For prompt-driven photoreal photo edits (object removal, keep subject / remove others), default to `nano-banana-2` via `generate_image_edit` — consult `list_models({ type: "image_editing" })`. Do not auto-pick Flux.
+**Kolbo MCP routing:** call `generate_image` or `generate_image_edit`. Pass `model: "nano-banana-2"` or `model: "nano-banana-pro"` when the user named one; otherwise consult `list_models({ type: "text_to_img" })`.
 
 ## CRITICAL Kolbo Platform Rules
 

package/skill/references/models/prompt-copilot.md

@@ -123,7 +123,6 @@ If during the conversation it becomes clear the user is actually working with on
 | User mentions / asks for | Switch to |
 |---|---|
 | Seedance / Seedance 2 / Bytedance video | `models/seedance.md` |
-| Seedance 2.5 | `models/seedance25.md` (also load `models/seedance.md`) |
 | GPT Image 2 / gpt-image-2 / OpenAI image | `models/gpt-image.md` |
 | Nano Banana / Gemini image / Gemini 3 Pro Image | `models/nano-banana.md` |
 | Veo / Veo 3 / Veo 3.1 / Google video | `models/veo.md` |

package/skill/references/models/seedance.md

@@ -12,10 +12,10 @@ Load this file when the user wants a **Seedance 2 / Seedance 2.0** (ByteDance) v
 ## Universal Rules (apply to EVERY Seedance prompt)
 
 - **First line ALWAYS declares shot structure**: total duration, shot count, aspect ratio. Example: `Total: 15s / 6 shots / 16:9`. Put it at the BOTTOM of the prompt too. For connected narrative sequences the proven phrasing is `N connected cinematic shots, 15 seconds total, 16:9, Multishot ON` — use it and keep `Multishot ON` for any multi-shot story.
-- **Then the Locked Intro** — `[GLOBAL LOOK]` / `[CAST]` / `[LOCATION]` — before any shot. A one-liner `same character throughout` is not a character lock.
-- **Order inside each shot**: Subject → Action → Camera → Constraints → (Audio/SFX if relevant). Do NOT restack GLOBAL LOOK style inside the shot.
-- **Prompt length**: simple single-idea pieces ~120–280 words. Locked-intro cinematic typically 400–900 words. Shorter than ~120 words = random output. The 8000-char cap below always wins.
-- **Shot count is user-directed, not capped at 3 or 6.** If the user asks for N shots, deliver exactly N shots in one prompt unless they explicitly ask to split it.
+- **Order inside each shot**: Subject → Action → Camera → Style → Constraints → (Audio/SFX if relevant).
+- **Prompt length**: aim for ~120–280 words TOTAL across all shots combined (not per shot). Shorter than ~120 words = random output. Longer risks the 8000-char cap below and makes the model forget the opening. For 6-shot prompts, keep each shot 1–2 tight sentences.
+- **Character lock**: if a character recurs, open with `same character throughout all shots` to stop identity drift.
+- **Max 3 shots per single-shot prompt; max 6 shots in a multi-shot montage.** More causes drift.
 - **Always describe at least one camera movement per shot.**
 - **Tell Seedance what the camera is NOT doing** (e.g. `no cuts, no zoom, natural head movement`) — this is what locks POV.
 - **Final prompt is always English**, wrapped in a copy-ready code block. Detect intent in any language and reply in the user's language, but the prompt itself is English.
@@ -26,42 +26,13 @@ Load this file when the user wants a **Seedance 2 / Seedance 2.0** (ByteDance) v
   - **Never** split into multiple prompts, multiple code blocks, or "part 1 / part 2" to evade the cap.
   - Before outputting, internally count the characters of the final prompt as a single string. If > 8000, rewrite tighter and re-count. Repeat until ≤ 8000. Only then show the user.
 
-## Locked Intro (DEFAULT for any multi-shot cinematic)
-
-After the Total line, every multi-shot prompt — and any piece with recurring people or a recurring place — opens with three locked blocks. Same camera, same people, same world in every cut. Do not skip them. Do not restack this look as per-shot style.
-
-Skip only for: true single-shot POV/orb, 3×3 grid-panel mode, or video-edit tasks. UGC/phone pieces still use the three headers, but GLOBAL LOOK is phone-native (no "cinematic", no film-camera body).
-
-```
-Total: Xs / N shots / AR
-N connected cinematic shots, Xs total, AR, Multishot ON
-
-[GLOBAL LOOK – LOCKED, APPLIES TO EVERY SHOT]
-Arricam LT, Cooke S4/i primes, 35mm Kodak Vision3 500T, 1.85:1 spherical, T2.8. Shallow depth of field, halation on highlights, fine organic grain, lifted milky blacks, low contrast, no sharpening, no HDR. [Named] grade: [hero tone], [field colors], [shadows], [bloom]. Handheld with micro-drift, never locked off. Naturalistic performance, real dialogue sync, no music.
-
-[CAST – IDENTICAL IN EVERY SHOT]
-NAME: age, build, hair, face, wardrobe, skin, signature details. @Tag when a reference or DNA exists.
-PROP: recurring object.
-
-[LOCATION]
-Place in materials + light + color field. Blocking: who sits/stands where. Background LIFE (extras, ambient motion).
-
-SHOT 1 — 0:00–0:02 — Medium / camera position
-(director-dictated action — physical verbs, timed acting, quoted dialogue)
-Total: Xs / N shots / AR
-```
-
-Adapt the GLOBAL LOOK package to THIS world — never paste a beach grade onto a night interior. Equipment names (body / lens / stock / stop) live in GLOBAL LOOK only. Never name a real director.
-
-**Acting:** shot size → physical action → line → timing. `laughs obnoxiously for half a second, then flat and certain` — unbounded laughs eat 2s. Background LIFE on every wide/medium or the frame reads as a ghost town.
-
 ## The 5 Formats
 
 ### 1. Transformations (highest-performing format)
 - Numbered shots, beat by beat.
 - Escalation arc: **calm → threat → transformation → aftermath**.
 - 6 shots / 15s / 16:9 is the proven structure.
-- Prefer the **Locked Intro** over a style dump. For a quick action montage with no recurring cast, a short GLOBAL LOOK still beats the old ARRI-prefix pile.
+- Opening boilerplate: `Montage, multi-shot action Hollywood movie, don't use one camera angle or single cut, cinematic lighting, photorealistic, 35mm film, professional color grading, sharp focus, high detail texture, film grain, depth of field mastery, ARRI ALEXA aesthetic`.
 - **Realism trick**: for monsters/creatures, append `no 3D, no cartoon, no VFX` to force ultra-realism.
 - **Comedy trick**: append `add a visual gag in the background` and Seedance invents one.
 
@@ -81,7 +52,7 @@ Adapt the GLOBAL LOOK package to THIS world — never paste a beach grade onto a
 - Always supply: **clear location, clear power mismatch, defined escalation arc**.
 - Describe choreography beat by beat — Seedance executes what you write.
 - Single continuous shot 15s works for two-fighter scenes; describe camera moves between beats (`crests rooftop edge`, `full 360 orbit`, `pulls back to wide`, `descends with them`).
-- Use speed-ramping with impact slow-motion as the style anchor when comedic/stylized.
+- Use `Guy Ritchie speed-ramping with Snyder impact slow-motion` as the style anchor when comedic/stylized.
 
 ### 5. Animation (3D stylized)
 - Break the 15s into **timed segments** (`0–3s`, `3–6s`, `6–9s`, `9–12s`, `12–15s`) and describe each explicitly.
@@ -93,7 +64,10 @@ Adapt the GLOBAL LOOK package to THIS world — never paste a beach grade onto a
 
 Use whenever the user gives named characters or multiple reference images (`@Image1`, `@Image2`, …) — a tactical unit clearing a bunker, a duel between two referenced characters, a war scene. **This is always an Elements-mode prompt** (route the card to `elements`). Structure:
 
-1. **Locked Intro FIRST** (GLOBAL LOOK / CAST / LOCATION) — CAST is one line per person: `Name @ImageN — wardrobe, position in frame, what they carry`. End CAST with "All must match their character references exactly." LOCATION carries time-of-day + materials + light direction.
+1. **Labeled scene header FIRST** (grounds the scene before any shot):
+   - `Time of day:` — hour + light quality + atmosphere (dust, haze, heavy silence before action).
+   - `Location:` — the environment in concrete physical detail (materials, wear, light direction, high-contrast blown-out entrance, etc.).
+   - `Characters:` — ONE line per person: `Name @ImageN — wardrobe, position in frame, what they carry`. End with "All must match their character references exactly."
 2. **REFERENCE CONSISTENCY block** — map every reference and pin what must NOT change: `Reference Image 1 is <X>. Preserve exact face, hair, anatomy, wardrobe, colors, props.` Add per-character energy/aura color rules, and any already-established story state (e.g. "the gem is already shattered — no intact gem, no red glow"). End with "Do not redesign, morph, recolor, or swap either character, their clothing, anatomy, weapons, or the environment."
 3. **Shots** — either titled (`Shot 1 — Medium Wide / Tactical Positioning`) or timecoded (`SHOT 1 — 0:00–0:03`); timecodes must sum to the total duration. Under each shot use **Camera → Action → Audio** in that order.
 4. **Continuity** — to chain a series, open with `Begin as a seamless continuation from <the exact last beat of the previous video>.`
@@ -128,9 +102,16 @@ The model reacts to what can be **seen and measured**, not to mood words. Transl
 - ❌ "tense scene" → ✅ "man freezes, slowly clenches his fist, light only from the side, half his face in shadow"
 - ❌ "cool cinematic shot of a car, epic, fast" → ✅ "low tracking shot alongside the car as it powers through a wet curve, headlights glowing, spray off the tyres, hard buffeting camera shake"
 
-### Style — LOCKED at the top, not redistributed
+### Style — DISTRIBUTED, not a prefix
+
+Never pile all style tokens at the top of the prompt. Each aspect lives in the block that already governs it:
 
-The GLOBAL LOOK block is the prefix. Per-shot style is ONLY what changes (shot size, move, height, this-shot light). Do not re-state film stock / lens / grade per shot. Do not pile a second style dump at the end.
+- Lighting → inside the shot's LIGHTING description
+- Lens / FOV → in OPTICS
+- Color → either an explicit grade (when strong / stylized) or folded into LOCATION + LIGHTING for naturalistic looks
+- Skin / acting → in PERFORMANCE
+- Physics → in PHYSICS
+- Format / resolution / grain → at the END as a suffix stack (before LOCKS)
 
 ### Shot sizes
 
@@ -170,7 +151,7 @@ Use only the discrete steps. Not "23°" — use 18° or 29°.
 - **Emotion through muscle movement**, not labels. ❌ "she looks sad" → ✅ "her eyes drop to the table, jaw tightens, she swallows once before answering."
 - **WB in Kelvin.** 3200K / 4000K / 5600K / 8500K. Pick ONE for the scene's mood.
 - **Color as material + light + role**, never a flat list. ❌ "she wears red, he wears blue" → ✅ "crimson silk scarf catching the cold tungsten spill from the corridor".
-- **Equipment names live in GLOBAL LOOK only** (body, lens, stock, stop). Never inside shots. **Never name a real director** and never write "in the style of [person]".
+- **No equipment names**, no director references, no "shot on ARRI / Sigma 85mm / Roger Deakins".
 
 ### Cuts and timing
 
@@ -193,19 +174,18 @@ Use only the discrete steps. Not "23°" — use 18° or 29°.
 
 ### Camera placement
 
-Place CAMERA in the **3rd position** of each shot's core layers (Subject → Action → Camera → Constraints). FOV gets ignored at the end, conflicts with identity at the front. GLOBAL LOOK already owns lens / stock / grade.
+Place CAMERA in the **3rd position** of each shot's core layers (Subject → Action → Camera → Style → Constraints). FOV gets ignored at the end, conflicts with identity at the front.
 
 ### Pre-flight checklist (before output)
 
-- Locked Intro present (`[GLOBAL LOOK]`, `[CAST]`, `[LOCATION]`) unless a documented skip?
-- GLOBAL LOOK not restated inside shots? Equipment names only there? No director names?
-- Acting timed? Background LIFE on wides/mediums?
+- Distributed style (no top-pile)?
 - One camera movement per time slice?
 - FOV in degrees from the table (not mm, not arbitrary)?
 - WB in Kelvin?
 - Speed in km/h, atmosphere in % or meters?
-- Color via material + light + role (grade named as law in GLOBAL LOOK)?
+- Color via material + light + role?
 - Positive phrasing (no "does not X")?
+- No equipment / director names?
 - Emotion through muscle, not labels?
 - Multishot: FOV per segment + "no drift mid-segment"?
 - 8000-char cap honored?

package/skill/references/models/seedance25.md

@@ -13,7 +13,7 @@ Load this file when the user wants a **Seedance 2.5** video (they said "2.5" / "
 
 - **Duration 4–30 seconds**, whole seconds. 30s IS supported.
 - **Up to 30 shots/cuts in ONE generation.** Deliver exactly N if N ≤ 30.
-- **Prompt cap 10,000 characters** for the entire prompt as one string.
+- **Prompt cap 30,000 characters** for the entire prompt as one string.
 - **Up to 50 reference medias / Visual DNA mentions** (`@Name`, `@ImageN`, `#Moodboard`). Every referenced asset must be tagged in the prompt text.
 - **Multimodal refs:** images + video clips + audio can all anchor one generation.
 
@@ -35,11 +35,11 @@ UGC / phone vertical: NEVER write "cinematic". GLOBAL LOOK is phone-native. Use
 
 ## Prompt length
 
-Simple ≤15s ~120–280 words. Locked-intro cinematic 15s typically 400–900 words. Full 30s / 15+ shots typically 700–1200 words / ~4k–9k chars. Hard cap 10,000. Never split into part 1 / part 2.
+Simple ≤15s ~120–280 words. Locked-intro cinematic 15s typically 400–900 words. Full 30s / 15+ shots typically 700–1200 words / ~4k–9k chars. Hard cap 30,000. Never split into part 1 / part 2.
 
 ## Feature-Block (optional, UNDER the Locked Intro)
 
-Reach for extra department passes only when the user wants "their best possible 30 seconds" AND the 10k budget still has room after GLOBAL LOOK / CAST / LOCATION. Never replace the Locked Intro.
+Reach for extra department passes only when the user wants "their best possible 30 seconds" AND the 30k budget still has room after GLOBAL LOOK / CAST / LOCATION. Never replace the Locked Intro.
 
 May add above GLOBAL LOOK: **EMOTIONAL INTENT** + **SIGNATURE MOMENT**.
 May add under the shot list: CAMERA timecode pass, SOUND timestamps, PHYSICS contract, EDITING/CONTINUITY, DIRECTORIAL NOTES.