Skip to content
FunCoding

Search

Search docs, Skills and MCP

launch-video

Make a showreel-grade motion-graphics launch video, with music, that introduces a CLI, library, or product from its repo. Use when asked for a launch, promo, sizzle, or showreel video, an animated intro for a project, or a video to post on X/social.

科研1kskills/graphics/launch-video/SKILL.md

Install

Send this to Claude Code, Codex or Cursor. The agent checks the Skill for safety first and installs it only after you confirm.

读取 https://funcoding.ai/skills/dzhng/skills/launch-video/install.md ,按里面的步骤帮我安装这个 Skill。

SKILL.md

Launch Video

The bar is a showreel: the piece a motion designer puts on their résumé to prove how good they are — dynamic, ambitious, scored to music, a real professional production and never a demo or prototype. Go all out. The repo supplies the story; the craft is yours. Nothing here is a template: if the piece could be re-skinned for any other project, it has failed.

Workflow

  1. Find the story. Read the README, docs, benchmarks, and brand art. Done when you can state, each with a source: what it is in one line, how it works underneath (the audience is technical), and the proof worth landing. Flag any claim the repo cannot back instead of inventing its meaning.
  2. Direct before you build. Choose one visual concept grown from what the project is (its metaphors, its brand art, its mechanism): a world with its own look, camera, and signature move, rather than a sequence of panels. Then commit to one familiar real-world analogy for the mechanism (building a city, running a kitchen, a heist) and map every step onto it: name each step in the analogy's vocabulary, and give each one the everyday props that explain it at a glance (a surveyor's tripod, stakes, cranes, an inspection stamp). Digestible beats clever: a viewer should get each step from references they already know, without reading a word. Write it as a shot list on a music timeline. Done when every shot names its camera move and the one moment in it a motion designer would be proud of. A shot whose description is "a card/panel/list appears" gets redesigned.
  3. Build picture and score from one timing source. Every event time lives in one place that both the scenes and the soundtrack read, so hits land on hits. Synthesizing the music in code keeps that lock exact.
  4. Review as a director. Render a draft, tile each shot into contact sheets, and get a fresh-eyes critique from a subagent that has not seen your plan. Then ask each shot the showreel question. Fix and re-render until every shot passes, not a sample.
  5. Ship. Make the thumbnail (rules below), render the master (1080p60, high-quality H.264, 320k AAC), reveal it in Finder, and commit the project to the repo as a standalone package, with renders and generated audio ignored and a short README covering commands, the timing principle, and the claim sources and methodology kept outside the film.

Marketing, not a white paper

Lead with the product and a few memorable benefits. Keep methodology, sample sizes, missing-data allowances, benchmark caveats and source citations in the package README or linked report, not in the video or key art. Do not add fine print, asterisk footnotes, disclaimer blocks or defensive narration. If a claim needs narrowing to stay true, narrow the headline itself; never use tiny text to rescue a misleading headline. Review each frame for reading burden as well as visual quality.

Show, don't tell

Text is the last resort. Before any word goes on screen, find the picture that makes it unnecessary: an object that transforms, a world that changes, motion that performs the idea. Keep text only where nothing visual can carry it: the product's real commands, its name, and a claim that must be exact. The one exception is orientation: give each shot one header saying what is happening (it can be dynamic, like a live count that ticks with the action), or the best animation reads as random motion. When a draft leans on labels, captions, or subtitles to explain what is happening, redesign the shot instead of rewording the text.

The showreel question

Would a motion designer put this shot on their reel? Tells that it has slid into explainer or demo:

  • Animated slide deck. A static layout where elements slide in and hold. A reel's camera is always going somewhere, and its composition keeps changing.
  • Reading, not watching. Paragraphs, UI cards, body copy, or a subtitle under a headline (a headline that needs a subtitle is two ideas). A reel speaks in few-word hits at dramatic scale, and the visuals carry the mechanism.
  • Broken metaphor. An action that doesn't do what it depicts: a cut that ignores the pieces it claims to divide, a label nobody could read against its ground. Make the motion true to the objects on screen.
  • One trick. Every element springs in the same way. A reel shows range: depth and parallax, scale contrast, match cuts and transitions that carry an element into the next shot, speed ramps, lighting and color shifts, particles and physics, typography that moves as design.
  • Flat. Everything sits on one plane at one scale, with no foreground, background, or light.
  • Deaf. Motion ignores the music, or the music sits under the picture instead of driving it.
  • Prototype 3D. Untextured primitives under flat light with no lens: it reads as a tech demo. If the piece goes 3D, finish it like a game cinematic: physically based materials with surface detail and bevels that catch light, reflections and soft shadows, emissives that bloom, and a lens (ambient occlusion, depth of field, grain, aberration on impacts).

Rules

  • Rebuild, never screenshot. Break any UI down into parts that can move.
  • Judge effects in a rendered frame, never the live preview. Frame-by-frame renderers (Remotion's manual advance, for one) can silently drop a post stack or a canvas layer that looks fine live.
  • Claims stay exact. Verify headline claims against the source and keep their scope clear in the headline. Preserve supporting evidence in the docs; follow the marketing rule above for what belongs on screen.
  • You cannot listen. Verify the mix by measurement (integrated loudness around −11 to −14 LUFS, and a spectrogram showing risers and impacts at their cue times), and say in the handoff that the audio was checked by measurement, not by ear.
  • Thumbnail. Feeds (X) pick an early frame and skip a one-frame poster, so hold the preview for ~0.3s of pre-roll, then move it into the opening (shift the music by the same amount). Match the pre-roll background to the opening so playback doesn't flash. The best preview is the product doing its job (name, tagline, a completed command and its output); the headline number gets its payoff shot and a separate key-art still.

Similar Skills

lead-research-assistant
ComposioHQ/awesome-claude-skills77k

lead-research-assistant

Identifies high-quality leads for your product or service by analyzing your business, searching for target companies, and providing actionable contact strategies. Perfect for sales, business development, and marketing professionals.

Science

13c-metabolic-flux
K-Dense-AI/scientific-agent-skills48k

13c-metabolic-flux

Estimates intracellular metabolic fluxes from steady-state carbon-13 isotope-tracing measurements using validated atom maps, mfapy isotope simulation, constrained multistart fitting, and flux-profile diagnostics. Use for 13C-MFA, carbon tracing, mass isotopomer distributions (MDVs/MIDs), positional isotopomers, parallel tracer experiments, and determining whether labeling data constrain a pathway flux. Distinguishes measured-label inference from COBRA flux balance analysis and flags experiments requiring nonstationary MFA.

Science

datamol
K-Dense-AI/scientific-agent-skills48k

datamol

Pythonic wrapper around RDKit with simplified interface and sensible defaults. Preferred for standard drug discovery including SMILES parsing, standardization, descriptors, fingerprints, clustering, 3D conformers, parallel processing. Returns native rdkit.Chem.Mol objects. For advanced control or custom parameters, use rdkit directly.

Science

biopython
K-Dense-AI/scientific-agent-skills48k

biopython

Provides Biopython workflows for sequence manipulation, file parsing (FASTA/GenBank/PDB), phylogenetics, and programmatic NCBI/PubMed access (Bio.Entrez). Supports batch processing, custom molecular-biology pipelines, BLAST automation, structure analysis, and motif analysis.

Science

bulk-rnaseq
K-Dense-AI/scientific-agent-skills48k

bulk-rnaseq

Prepares bulk RNA-seq FASTQ, Salmon, STAR or featureCounts output for gene-level differential expression. Covers nf-core/rnaseq and standalone quantification, biological replication, strandedness, reference provenance, validated count assembly and a PyDESeq2 handoff. Use for FASTQ-to-counts analysis, nf-core/rnaseq configuration, STAR/Salmon quantification, or building a counts matrix for DESeq2. For single-cell data use scanpy; for statistical fitting alone use pydeseq2.

Science

alphagenome
K-Dense-AI/scientific-agent-skills48k

alphagenome

Looks up precomputed AlphaGenome Atlas effects for any GRCh38 single-nucleotide variant (AVI score with Phred and 18 SHAP feature attributions, plus raw and quantile scores for RNA-seq, DNase, ATAC, ChIP-TF, ChIP-histone, CAGE, PRO-cap, splicing, polyadenylation and contact-map tracks), scores variants or scans windows on demand with the AlphaGenome model for human and mouse (variant scoring, in silico mutagenesis, REF-versus-ALT track prediction), and builds Atlas website deep links. Use when the user mentions AlphaGenome, AlphaGenome Atlas, AVI or AlphaGenome Variant Impact, DeepMind variant effect prediction, or wants to prioritise or mechanistically interpret non-coding, regulatory, splicing, enhancer, promoter, or chromatin-accessibility effects of SNVs from a VCF, credible set, or region. Research use only; not a clinical tool.

Science