All authors
Claude Skills by teen-boop
github.com/teen-boop34 skills0 installs19 views
- Ai Visual Production DirectorThe entry-point orchestrator for this whole skill bundle — turning a story/script into a consistent, identity-locked sequence of AI-generated images and video. Reads what phase of work the user is asking for (raw screenplay parsing, character/reference setup, turning prose into a shot list, shot sizing and continuity, reference assembly, still-image prompt writing, camera/transition tracking, video-prompt writing) and routes to the correct companion skill(s) in the correct order, instead of l...Votes: 0GitHub stars: 2
- Animation Effects CatalogGrowing catalog of proven tricks that make AI video (Seedance 2.x, Higgsfield, Kling, Veo) look cinematic and impressive — organised by HERO TYPE (tiny creatures, giants vs tiny humans, real animals, surreal anatomy, painted/puppet/material characters, levitation, falling/action humans, face recast) and by EFFECT (camera, physics and contact, performance, voice and sound, light and integration, workflow modes, negative blocks). Each entry gives when to use it and a copy-ready English phrase. ...Votes: 0GitHub stars: 2
- Character Identity Lock SetupBuilds a locked visual-identity reference system for a recurring character — a set of reference images (a multi-panel turnaround sheet plus a standalone bust portrait) for every distinct age or life-stage that character appears at, generated from real photo identity sources and hosted so every later generation can attach the correct one. Prevents a character's face from drifting between shots, and prevents an AI image generator from ever inventing the face freehand from text alone. Use at the...Votes: 0GitHub stars: 2
- Cinema Director V3Cinema director v3 for Seedance 2.0/2.5 and Higgsfield video prompts. Supersedes earlier cinema-director versions — prefer this one. Establishes the target Seedance version first, since 2.0 caps at 9 image references and 15 seconds while 2.5 allows 50 and 30. Writes production-grade multi-shot prompts on a locked 16-slot spine: shot header, Style Prefix, NO ON-SCREEN TEXT, CRITICAL blocks, Assets, Geometry Map, First Frame, lens locks in FOV degrees, Camera, Light and Colour, Atmosphere, time...Votes: 0GitHub stars: 2
- Cinematic Prompt WriterWrites manual, copy-paste-ready image-generation prompts in a locked period-cinematic style (anamorphic lens, Canon 5D Mark II, gritty premium theatrical grade). Given a scene/subject description, produces THREE framing variants — wide, medium, and close-up — all sharing the same subject and the same style block. Works across any project: if the current project has its own established style bible, character references, or QC rules, this skill picks them up and layers them on top of the base s...Votes: 0GitHub stars: 2
- Detail Reference IntakeWatches a project's "working detailed references" folder for new screenshots/stills captured from test generations or test animations, identifies what specific reusable detail each new file shows, and locks it into the project's canonical reference index so every later generation stage (storyboard, full panels, video) uses it. Use at the start of any work session on a project that has this folder, before storyboarding or generating new panels. Triggers on working detailed references, new refe...Votes: 0GitHub stars: 2
- Generation Backend SetupFirst step before any image or video generation in a film project. Asks the user which generation backend they can connect — a command-line tool (CLI), an HTTP API, an MCP connector already available in Claude, or nothing — and connects it safely: keys stay in environment variables or the OS keychain and are never written to project files, printed or committed. It tests the connection with the smallest possible call, sets the parallel-job limit, and records the choice (names of env vars only,...Votes: 0GitHub stars: 2
- Location Room MapTurns any location reference (usually one photo of one wall) into a COMPLETE, physically plausible room — a top-down floor plan, all four walls described and generated as separate wall references, every door/window/furniture piece fixed in place — so reverse shots, over-the-shoulder shots and close-ups always show the CORRECT wall behind the subject instead of repeating the one wall from the photo. Enforces architecture rules (windows on at most 2 walls, at least one door, one of each recurri...Votes: 0GitHub stars: 2
- Loveart Project DirectorThe entry-point orchestrator for a Love Art video-animation project. Reads what phase of work the user is asking for (project setup, reference locking, storyboarding, full-panel generation, video-prompt writing) and routes to the correct companion skill(s) in the correct order, instead of leaving skill sequencing to memory. Use whenever starting or resuming work on a Love Art story/animation project, or when it's unclear which skill/stage applies to the current request. Triggers on new projec...Votes: 0GitHub stars: 2
- Loveart Video AnimationSkill for creating consistent image sequences for video animation exclusively in Love Art. Use when user works with story scripts, reference images, character face replacement while preserving style lighting and pose, location databases, recurring subjects/objects (pets, secondary characters, props, vehicles, etc.), clothing consistency, multi-shot adaptation from existing wide shot to close-up inside same location, Nano Banana 2 or GPT Image 2.0 generations. Triggers on video animation, love...Votes: 0GitHub stars: 2
- Manual Image PackHand-generation workflow for image frames — the image counterpart of the video-prompt pipeline. When automated generation is slow, blocked or too expensive, this skill turns every frame still to be made into a ready-to-use folder: PROMPT.txt to copy, the reference images already copied in and numbered in attach order, INFO.txt with what/how many/where, and a slot for the result (RESULT_1.png …). Then it collects the user's finished frames back into the project (format check, auto-crop of thin...Votes: 0GitHub stars: 2
- Motion Design ReferencesReference and design intake for a motion-graphics project — the step between locking a STYLE PACK and generating anything. Inspects what is actually in the project's references folder first and reports it, then has the user assign roles; sorts every reference into STYLE references (which teach the look and feed the pack) versus ASSET references (which supply a specific thing), builds the folder taxonomy and an INDEX, and defines which references get attached to which generation in Lovart. Eac...Votes: 0GitHub stars: 2
- Motion Graphics DirectorEntry-point orchestrator for AI motion-graphics work (Vox-style explainers, animated charts, collage clips). Decides which of the two production routes applies — CODE (Remotion + Claude Code, for numbers/charts/typography/programmatic scale) or VIDEO (Omni Flash / video models, for dynamic collage, camera moves, big blocks) — then routes to the correct companion skill in the correct order. Owns the shared prompt library of 46 composable animation blocks. Use whenever starting or resuming a mo...Votes: 0GitHub stars: 2
- Motion Remotion BuildThe CODE route for motion graphics — building animated charts, editorial explainers and Vox-style scenes in Remotion driven by Claude Code, with no After Effects and no keyframes. Covers the one-prompt project setup, the exact-numbers doctrine, determinism rules that stop render flicker, prop controls and saving them back, the three-layer scene architecture, master sequence assembly, voice-over sync, QA toggles and render. Use whenever building or fixing a Remotion motion-graphics composition...Votes: 0GitHub stars: 2
- Motion Style CoreBuilds and locks a per-project STYLE PACK for motion graphics — the swappable design layer. Every project gets its own pack (fonts, palette with roles, surface, edge treatment, depth, motion vocabulary, negatives) filled against one fixed schema, so the pipeline stays constant while the design changes completely from project to project. Derives a new pack from a reference image, a brand, or a described mood; ships tested packs (paper-vox, blueprint, clay) as starting points, never as defaults...Votes: 0GitHub stars: 2
- Motion Video RouteThe VIDEO route for motion graphics — producing animated collage/explainer clips with video models (Gemini Omni Flash first, Seedance as fallback) from a locked master style sheet. Covers the safe route (style ref → image → video) versus the fast route (style ref + prompt → video), the image-prompt and video-prompt templates, the mandatory style restatement inside video prompts, the big-blocks rule of thumb, the known-faces workaround, and model routing. Use when a shot wants dynamic camera, ...Votes: 0GitHub stars: 2
- Scene Continuity LockPrompt-level continuity guard for one scene — for BOTH storyboard panels and final frames. Builds a SCENE STATE sheet (who sits/lies where on which piece of furniture, pose, held props, prop states such as a box open or closed, light, time, camera axis) and checks every shot prompt against the PREVIOUS and the NEXT prompt of the same scene before generation — then injects a verbatim CONTINUITY block into each prompt and QCs each generated frame against it. Catches heroes changing seats, a cha...Votes: 0GitHub stars: 2
- Script ExtractorParses source material — a formatted screenplay (sluglines, character cues, action lines) OR any free narrative text (a short story, a novel excerpt, a literary adaptation, prose fiction with dialogue) — scene by scene (or scene-equivalent unit, for prose with no formal scene markers), and extracts five categories of detail per unit: locations (INT/EXT, day/night, inferring the time-of-day when it isn't explicitly stated), characters present, props, key physical actions, and wardrobe/appearan...Votes: 0GitHub stars: 2
- Seedance Prompt WriterWrites production-grade Seedance 2.0/2.5 video prompts on a locked 16-slot spine (header, style prefix, no-on-screen-text, CRITICAL blocks, assets, geometry map, first frame, FOV-degree optics, camera, action timing, physics chain, acting, audio, locks), merged with this project's mandatory rules: NO MUSIC, English-by-default, every prompt fully standalone (no citations to other clips), percentage-based color doctrine tied to a physical source, and shot-size reasoning from storyboard-shot-sel...Votes: 0GitHub stars: 2
- Seedance Shotlist TrackerPackages an already-written set of Seedance prompts (from seedance-prompt-writer) into ONE self-contained, interactive HTML shot-list document — a checkbox per scene/clip, a copy button on every prompt, progress persisted in the viewer's browser. This is a delivery/ tracking layer, not a prompt-writing method: it never invents its own prompt content or a weaker style-prefix format, it packages the project's real, already-approved prompts verbatim. Use after a batch of clips for a scene or epi...Votes: 0GitHub stars: 2
- Storyboard Camera Continuity LedgerReads a sequence's narrative/shot list, extracts the camera movement within each shot and the transition type between consecutive shots (push-in, pull-back, pan, tilt, crane, tracking, whip, match-cut, straight cut, etc.), and persists them into a per-project continuity ledger file — a small database other work (animation/video prompts, editing, future sequences reusing the same locations) can read back later instead of re-deriving camera logic from scratch. Also cross-checks that image/video...Votes: 0GitHub stars: 2
- Storyboard Continuity TrackerTracks the physical state (position, pose, action) of the character, every recurring subject/object, and actively-handled props across a sequence of storyboard shots, and validates that each shot's starting state either matches the previous shot's ending state or is an explicit, shown, logical consequence of it. Use when building or ordering a multi-shot storyboard/raskadrovka, checking whether shot N+1 logically follows from shot N, or diagnosing why a shot sequence feels disconnected. Trigg...Votes: 0GitHub stars: 2
- Storyboard Narrative BreakdownTakes a continuous, unstructured narrative description of a sequence (a paragraph of prose describing a camera moving through a series of events and locations, in any language) and breaks it into a numbered list of discrete mini-scenes, each tagged with its location, its event/subject, and the visual element (if any) that bridges it to the previous and next mini-scene. For each mini-scene, writes the actual generation prompt(s) (via `cinematic-prompt-writer` or the current project's own promp...Votes: 0GitHub stars: 2
- Storyboard Reference AssemblyAssembles and verifies the complete set of locked reference images (character, recurring subjects/objects, exact location angle, props/details) that must be attached to a storyboard panel or shot before it is generated. Use when building a storyboard/raskadrovka, writing per-panel or per-shot prompts, deciding which reference images to attach, or checking whether a generated panel actually matches the project's canonical assets. Triggers on storyboard panel, raskadrovka, which references to a...Votes: 0GitHub stars: 2
- Storyboard Shot SelectionDetermines, for every beat in a script, WHY a specific shot size/type (establishing, wide, full, cowboy, medium, medium close-up, close-up, extreme close-up, group shot, over-the-shoulder, POV, dutch angle) is the correct cinematographic choice for that beat, and writes that reason directly into the storyboard shot description. Prevents two adjacent shots in a sequence from sharing the same underlying purpose or framing without narrative justification. Use when building a storyboard/raskadrov...Votes: 0GitHub stars: 2
- Video Clip Continuity ChainPrevents prop/subject/object state mismatches at the boundary between separately-generated video clips (e.g. Seedance) that are meant to play back to back as one continuous scene. Forces every clip's prompt to explicitly restate the exact inherited state of every tracked entity from the end of the previous clip, instead of leaving it implied. Use whenever writing a video-generation prompt that continues from an already-written or already-generated clip, splitting a scene into multiple video p...Votes: 0GitHub stars: 2
- Visual Story 1 SetupSTEP 1 of 8 in the sequential visual-story pipeline. Project setup for turning a story into a consistent, identity-locked sequence of AI-generated images/video. Establishes the working folder, resolution, aspect ratio, project type, the references folder hierarchy (Characters / Locations / Props), Story-agnostic: FIRST inspect the references folder and report what is actually there, THEN ask the user to assign roles (which reference is the main character, which is secondary, the rest — people...Votes: 0GitHub stars: 2
- Visual Story 2 ScriptSTEP 2 of 8 in the sequential visual-story pipeline. Turns a story idea (or an existing draft) into a finished, structured script using industry-standard structures (Save the Cat / Hero's Journey / Three-Act / Sequence Approach). Includes ready-to-use Russian screenwriting prompt templates. Run AFTER visual-story-1-setup (project setup). Next step: visual-story-3-cast-world.Votes: 0GitHub stars: 2
- Visual Story 3 Cast WorldSTEP 3 of 8 in the sequential visual-story pipeline. Cast & world development — the planning pass done BEFORE generating any reference sheet. Go character by character (people AND animals) and define each one's goal/tasks and arc, costume option(s), which locations they appear in, and which supplied references fit them; then enumerate every location and every recurring prop the story needs. The number of characters is driven by how many real references exist. Output is a written cast/world bi...Votes: 0GitHub stars: 2
- Visual Story 4 ReferencesSTEP 4 of 8 in the sequential visual-story pipeline. The reference-building step. For each scene, extract every character, location, and recurring prop from the scene's prompt, and generate a locked reference sheet for each one using a single shared 4-part template (character / location / prop sheets all share the same structure). This is what keeps people, rooms, and objects consistent across scenes. Run AFTER visual-story-3-cast-world. Next step: visual-story-5-exposition.Votes: 0GitHub stars: 2
- Visual Story 5 ExpositionSTEP 5 of 8 in the sequential visual-story pipeline. Exposition / establishing shots. For each location, generate the wide establishing shot that opens its scenes and sets up the world (space, scale, palette, light, era) — the shot that "exposes" the location before any character coverage. Each establishing shot doubles as that location's master reference for the scene, attached to every subsequent shot there. Built from the locked location sheets, obeys the STYLE LOCK block. Run AFTER visual...Votes: 0GitHub stars: 2
- Visual Story 6 ProductionSTEP 6 of 8 in the sequential visual-story pipeline. Scene production. Builds the STYLE LOCK block, establishes master shots, generates every mini-scene's coverage attaching the locked character/location/prop references, uses the "Change to..." technique for sequential shots, and propagates continuity/weather across a scene. Run AFTER visual-story-5-exposition (reference sheets built). Next step: visual-story-7-platform-qc.Votes: 0GitHub stars: 2
- Visual Story 7 Platform QcSTEP 7 of 8 in the sequential visual-story pipeline. Platform operation and quality control. Covers platform-specific mechanics for Lovart and Krea (auth, free/paid mode, concurrency limits, timeout/poll behavior, aspect-ratio handling, real-face-upload restriction on video), the budget/mode confirmation rule, and the QC + identity-fix pass before delivery. Run AFTER visual-story-6-production, and reference its platform notes throughout generation. Runs alongside generation; the pipeline's fi...Votes: 0GitHub stars: 2
- Visual Story 8 StoryboardFINAL production step of the sequential visual-story pipeline — the storyboard. This is what actually goes into production, done ONLY after the script is final, the Production Design Guide is locked, the scene stills are face-locked, and all reference tests have passed. Produces a shot list (table), storyboard frames (numbered panels with slug-line headers + ACTION/DIALOGUE captions), and a motion-annotated layer (RED body / BLUE camera / GREEN framing / ORANGE lighting / YELLOW VFX) that dec...Votes: 0GitHub stars: 2