aristo — AI video skills + tools in one place.
skills from real AI creators. every top model. your own agent. one workspace.
start with aristo →skills
skills let you make anything — from the best creators.
a skill is a playbook: your agent loads it and walks you through the author's process — prompts, techniques, workflow.
ultra-real 'found home video' clips: subject/location/style/camera blocks + a per-second timeline + ambient-only audio, in an early-2000s DV camcorder look
@john_my07
generate the dialogue audio first, validate the performance for cents, then drive the video to sync with it — better acting, far cheaper iteration
@TomLikesRobots
exact camera moves for action scenes: block the scene in Blender, render a rough previs clip, feed it to seedance-video as a motion reference
@reidhannaford
turn one viral TikTok into a reusable beat-by-beat template, then batch-produce UGC variants from your own reaction clips and product demos
@danielhangan_
the most reliable low-budget AI ad format: a genuine-looking reaction clip + a 5-7s product demo, produced in volume and posted daily
@maubaron
never ask for 'a landing page' — name a design language (Glassmorphism SaaS, Neo-Brutalist, Luxury Editorial...) plus palette, type, layout and motion rules
@aiwithmayank
one event, two synchronized perspectives in a single generation: two equal vertical halves, frame-synced, filmed by two different cameras
@umesh_ai
structure a full 30s film scene as one prompt: continuous camera choreography, dialogue beats and blocking mapped to timecodes — feature-film workflow
@PJaccetturo
handheld feel you can't prompt: record a real camera move by walking through a space with your phone, then transfer it as a motion reference
@maxprokopp
video and audio reference limits are separate (15s each) — slice one master soundtrack into per-shot audio refs so sound flows continuously across cuts
@rainisto
for music videos, don't script individual scenes — prompt the mood, pacing, animation style, camera language and story arc, and let visuals evolve with the track
@john_my07
Direct a scene as a Three.js previz — staging, cameras, edit — then have seedance-25 shoot it for real: schematic figures, number tags, a location plate born from the previz frame. Cuts transfer frame-accurately. (by Michael K)
M Michael Ktools
every top model, one balance.
- video — seedance, minimax h3, veo 3.1, kling, gemini omni
- image — gpt image 2, nano banana, seedream 5
- plus upscale, speech, music, transcription, code sandbox
your agent + workspace
- connect your own agent — chatgpt, claude or any mcp client, your choice
- all assets and generations live in your workspace