Assemble the FREE steps of the flat-vector-explainer video format — a flat-illustration host character (who, where, tone, voice and music come from the recipe choices) walks a countable N-step product routine, one step per beat, and Remotion composites every chip/numeral/tagline/slate/CTA as an animated DOM overlay ON TOP of the Kling i2v character clips (text is NEVER baked into a keyframe — i2v warps type), the closing 'N products' grid is a PIL composite of the REAL product photos (not AI), full-sentence VO drives word-by-word burned captions over a VO-forward music bed, and the ~50s animated silent master is re-cut to a 30s deliverable FROM the animated master (never a static intermediate). Documentation-grade — ships config.example.json + PIPELINE.md + a README of the free assembly; the paid gen steps (keyframes, Kling i2v, VO, music) are separate capabilities the recipe orchestrates. Use for the flat-vector-explainer format.
Install gooseworks: https://gooseworks.ai/creative-agent/install.md # then, in Claude Code, Cursor, or Codex: /gooseworks use the render-flat-vector-explainer skill
Assembles a flat-vector product-routine explainer: one illustrated host character walks through a countable N-step routine (e.g. the demo's soil mix -> plant food -> leaf mist -> root tonic), one step per beat, each beat carrying a large corner numeral, a labelled chip + one-line tagline, and the step's real product photo, closing on an "N products" grid + brand CTA. It reads as a premium DTC explainer (Spotify/Anchor flat-vector lineage), not UGC.
This capability is documentation-grade. The format's source recipe is a documented recipe, not a runnable end-to-end app, so this capability ships the config schema (scripts/config.example.json), the field-to-script map (scripts/PIPELINE.md), and a README (scripts/README.md) describing the FREE assembly steps the agent runs by hand with ffmpeg + Remotion + PIL. The paid generative steps are separate capabilities the recipe orchestrates and gates.
Creative calls the user makes (the recipe's choices, asked in one round before any paid step). The demo's value is only an example, never the default. The flat-vector 2D style itself is the format, not a choice.
host — who the illustrated host is (gender, age, look, outfit) -> character.anchor_prompt. Asked; the demo used a mid-30s woman with a short black bob in green overalls.setting — where the routine happens -> the character keyframe prompts. Asked; the demo used a sunny windowsill full of houseplants.tone — the script's tone -> VO lines + expressions. Asked; the demo used faintly comedic overwhelm -> satisfied payoff.narrator_voice — the VO voice -> voice.voice_id. Asked; the demo used ElevenLabs "Eryn".music — the bed -> music.prompt (or none). Asked; the demo used lo-fi pop, 95-105 BPM.scripts/config.example.json is the Greenmoth worked example: copy its structure, never its creative values or its brand palette.
The agent runs these deterministic, free steps by hand — see scripts/README.md for the ffmpeg/Remotion/PIL detail:
loudnorm I=-15 VO-forward, mux, burn captions LAST -> finals/master-final.mp4 (~50s).finals/master-final-30s-v1.mp4.The recipe orchestrates and gates these; they are not part of this capability:
create-image-fal (nano-banana; re-render a FRESH flat-vector anchor, never chain a photoreal ref).create-video-fal (Kling v3 standard image-to-video with start_image_url and generate_audio: false, cfg 0.5; Kling 2.5 Turbo is withdrawn, style-preserving negative, low motion; TEST one scene before batching).create-vo-elevenlabs (eleven_v3, with-timestamps).create-music-elevenlabs.host* — who the illustrated host is (gender, age, look, outfit) -> character.anchor_prompt. Asked; the demo used a mid-30s woman with a short black bob in green overalls.setting* — where the routine happens -> the character keyframe prompts. Asked; the demo used a sunny windowsill full of houseplants.tone* — the script's tone -> VO lines + expressions. Asked; the demo used faintly comedic overwhelm -> satisfied payoff.narrator_voice* — the VO voice -> voice.voice_id. Asked; the demo used ElevenLabs "Eryn".music* — the bed -> music.prompt (or none). Asked; the demo used lo-fi pop, 95-105 BPM.Research supported advertising promises from a specific buyer's context, current product facts, customer language and observed ads. Turn them into static, video or copy ideas; match video ideas to actual recipes and save a product-scoped angle bank for write-video-ad-script. Product-led discovery can use verified facts without reviews. Reference persistence and reach are observations, not proof of performance.
Analyze ad campaign performance data (Google, Meta, LinkedIn) to identify what's working, what's wasting budget, and specific cut/scale/test recommendations. Runs statistical analysis, funnel diagnostics, and multi-channel budget reallocation with specific dollar-amount shift recommendations and scenario modeling.
For paid lead-gen and participant-recruitment ads, replaces vanity CPA with true CAC per qualified lead by joining ad-platform data with downstream funnel events, surfaces tracking gaps, and classifies every creative into Scale / Keep / Investigate / Cut.
A seven-step workflow for turning a concept brief into a finished animated explainer ad with Claude Code, image models and ffmpeg.