Claude Skills by manishiitg

Back up the family's learning workspace to the parent's own private Hugging Face Hub dataset repo — the ONE destination this family uses, via the `hf` CLI.

Build a designed, self-contained HTML academic map of the child's subjects, topics, and materials, plus real coaching notes for the parent on how to teach them better.

Build a designed, self-contained HTML progress report for the child — how she's actually doing, and what to work on next — surfaced in the Progress tab for both parent and child.

Create clear, child-ready study material (notes, worked examples, a revision sheet) for a topic, matched to the child's level and their own uploaded materials, packaged as one activity.

Create a practice test for the child from their materials and progress — child-facing questions with a separate parent-only answer key, packaged as one activity.

A fun, off-syllabus "something new to explore" activity for the child — a light cover page plus a chat-driven discovery, tailored to grade and interests learned over time. Parent-initiated, handed off like any other activity.

Send the parent a notification at the right moments — the child finished a session, a test/report is ready, a backup completed, or a decision is needed.

Read a newly uploaded file from the inbox, classify it, move it into the right subject/topic folder, and write a metadata JSON.

Publish a family artifact (a progress report or the academic map) to a shareable destination — a shared folder, a static host, Google Drive, or GitHub Pages — tracking config vs status in files.

Extract the actual text/content from a file of ANY format — PDF, Word, PowerPoint, Excel, images, scans, archives. The canonical "how do I read this file" reference, reused by process-file, Pulse ingestion, and browser downloads.

Age/grade-appropriate approach for a coding/programming subject — read this BEFORE create-study-material or create-test when the topic is coding/programming, in addition to that skill.

Extract genuine interest signals from the child's own conversations and keep memory/interests.md current — a Pulse-only skill, not triggered by a direct parent request.

Extract durable things the parent has said in chat and keep memory/preferences.md current — a Pulse-only skill, not triggered by a direct parent request.

Create and present the user-approved visual evidence that a cinematic sequence needs before video generation: FLUX.2 Max character/background masters and controlled derivatives, subject reference packs, locations, wardrobe, props, lighting, camera language, and start/exit/bridge frames. Use before an anchor or follow-up clip.

Generate AI video, image, voice/TTS, and music clips via fal.ai's Node.js client for cinematic video productions. Use when a brief calls for footage, reference imagery, narration, or a music bed that cannot come from uploaded assets or a selective deterministic HyperFrames insert.

Plan and generate or edit video effectively with Google Gemini Omni Flash through current fal.ai endpoints. Use when a production selects or considers Gemini Omni, Google Omni, or Gemini Omni Flash for text-to-video, image-to-video, reference-to-video, conversational video editing, localized revisions, multimodal conditioning, native audio, or iterative scene continuity. Read with video-provider-capabilities and fal-ai before any paid call.

Check AI-generated footage for the failure modes generation causes rather than deterministic editing -- identity drift, generation artifacts, motion that breaks physics, temporal discontinuity at stitch points, lip-sync drift, and prompt adherence. Use alongside video-quality for any candidate assembled from fal-ai or google-ai clips; never in place of it.

Generate optional still images and off-camera narration through Google's Gemini API for Video Studio. Do not use Google video models: all Video Studio footage uses MiniMax H3 through fal.ai.

Design and render video frames as HTML/CSS, then assemble them into a video with headless Chrome and ffmpeg. Use when building an infographic, product explainer, stat card, title sequence, or any piece whose visuals are laid-out typography, charts, and shapes rather than photographic footage.

Review and gate an editable HyperFrames composition or its rendered video before delivery. Use after building, revising, diagnosing, or rendering a HyperFrames project when layout, contrast, deterministic timing, transitions, animation choreography, rendered-frame fidelity, or high-quality acceptance needs verification. Use alongside video-quality; this skill owns composition-specific checks, while video-quality owns final media, editorial, audio, claims, and presentation QA.

Plan and generate Kling video effectively through current fal.ai Kling endpoints. Use when a production selects or considers Kling for text-to-video, image-to-video, reference-to-video, first/last-frame interpolation, structured multi-shot prompts, recurring characters or objects, style references, motion transfer, or native audio. Read with video-provider-capabilities and fal-ai before any paid Kling call.

Direct a coherent cinematic video of any duration from story architecture through deterministic or generated sequences, continuity-controlled clips, editorial stitching, sound, and final seam review. Use as the default director for every new Video Studio production so scenes feel like one continuous authored film rather than separate generations.

Plan and generate Video Studio's MiniMax H3 Max routes through fal.ai: Text-to-Video for prompt-only shots, Image-to-Video for approved first/last-frame control, and Reference-to-Video for identity and continuity. Read with video-provider-capabilities and fal-ai before any paid H3 Max call.

Design a coherent MiniMax H3 sequence using a reference manifest and explicit camera-transition grammar. Use before planning or creating an anchor clip or any follow-up clip in a cinematic sequence.

Plan and generate Seedance video effectively through current Seedance 2.0 and 2.5 endpoints on fal.ai, or supported direct Seeddance models through its separately authenticated API. Use when a production selects or considers Seedance for text-to-video, image-to-video, first/last-frame control, multimodal reference-to-video, multiple character or product images, reference video or audio, native synchronized audio, multi-shot scenes, editing, extension, or fewer longer clips. Read with video-pr...

Generate supported Seedance video through the independent Seeddance API at seeddance.io using the deployment's SEEDANCE_API_KEY, separately from fal.ai. Use when choosing between direct Seeddance and fal, checking direct account credits or model access, or submitting, polling, downloading, retrying, and reviewing direct Seedance video tasks. Read with seedance-video, video-provider-capabilities, and video-cinematography before paid generation.

Plan and generate Google Veo video effectively through current fal.ai, Gemini API, or Vertex AI contracts. Use when a production selects or considers Veo for text-to-video, image-to-video, first/last-frame interpolation, reference-image identity or product guidance, Veo video extension, portrait video, high resolution, native audio, or long-running generation operations. Read with video-provider-capabilities and the selected provider skill before any paid Veo call.

Construct the actual MiniMax H3 video prompt -- the five-aspect formula (subject, motion, scene, spatial framing, camera), precise camera vocabulary, and character consistency across shots. Use after video-storytelling has placed a beat and whenever framing, motion, lighting, or identity drifted from the brief.

Direct a video project from a conversational brief through a reproducible production. Use when creating a new video, deciding the production approach, turning uploaded assets into a video, scripting or storyboarding, or revising an existing video.

Edit and assemble video clips with deterministic local tools, including trimming, normalization, hard cuts, captions, branded overlays, narration, music ducking, and versioned exports. Use when combining clips or uploaded media, adding captions or audio, resizing for a platform, or revising an existing edit -- including stitching multiple independently AI-generated clips (fal-ai, google-ai) into one long-form piece when each clip's duration doesn't match the storyboard beat.

Lock the visual world and complete audio direction before narration or footage generation. Use when deciding locations, backgrounds, wardrobe, lighting, palette, narrator or dialogue voice, music, ambience, sound effects, captions, and whether speech should be separate TTS or native model audio.

Choose which fal.ai or Google Gemini/Veo model fits one shot's requirements -- text-to-video vs image-to-video, character/subject consistency, duration limits, native audio, cost. Use before the first generation call in a long-form production, and again whenever a shot's requirements differ from the ones already resolved. Read alongside fal-ai and google-ai, which own the client mechanics once a model is chosen; this skill owns the choice itself.

Resolve an AI-video model's live official API schema and turn it into a durable request, continuity, cost, and review plan before any paid video call. Use for Kling, Seedance, Veo, or any other video model when selecting a model, preparing or retrying a request, assigning image/video/audio references, using first or last frames, creating multi-shot video, extending footage, chaining clips, or deciding whether separate clips are necessary.

Validate a candidate video technically, visually, and editorially before presenting it as complete. Use after rendering a video or when diagnosing playback, aspect ratio, duration, audio, captions, black frames, freezes, identity drift, or other quality problems.

Assemble an H3 Max continuity chain with a plain deterministic concat. Use for final assembly, or only when an H3 continuation has a visibly broken boundary.

Structure a video's narrative arc and pacing, scaled to its actual duration -- from a short explainer's single arc to a true long-form (8+ minute) piece's chapters, retention curve, and pattern interrupts. Use once the brief's message and audience are confirmed and before writing a shot-by-shot storyboard. Read alongside video-cinematography, which owns per-shot camera/lighting direction once a beat is chosen.