agentleFS
neskuchnyneskuchny/content_factory/.claude/skills/ai-visualization/SKILL.md

ai-visualization

Skill1 starsChanged 2 days agoMITView on GitHub

Create \"what it could look like\" AI visualizations through the Higgsfield MCP: the client's real product on a site or in an interior, a structure appearing and being assembled, lifestyle scenes. Preserves the product's geometry and always carries a label. Use for content units of type V and for pins that use AI.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before steps 1 and 3, estimate the credits and record them in weeks/<W>/costs.md.

    line 32

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AI visualization of a real product

Full flows: flows/f04-object-insert-broll.md (insert as a separate segment), flows/f05-object-insert-inshot.md (inserted directly into a shot with a person). For seams, see the seamless-insert skill.

Input

  • The unit's script (what to show, in what order).
  • Product references from product-base/media/: real photos of the product from 2–3 angles, dimensions, profile color. Do not start without references; request them through the orchestrator.
  • A photo of the site or house (if the scene is "at the client's place") or a description of the surroundings.

Pipeline (cheap first, expensive later)

  1. Key frames (images). generate_image with an edit model (Nano Banana Pro / GPT Image / Seedream) and the product references. You need 2–4 frames of the scene: empty site → structure → life in it (evening, dinner, plants). Prompt: product, angle, light, time of day, surroundings. The prompt must end with: "preserve the geometry, profiles and proportions of the product exactly as in the reference; no text in the frame".
  2. Frame review (reviewer). Rejects are filtered out here, while it is still cheap.
  3. Video. generate_video image-to-video or start/end frame (Kling / Seedance / Cinema Studio). One camera move per clip, 5–8 s. The "structure appears" effect is a transition from the empty frame to the finished frame.
  4. Video review (reviewer): posts, glazing, flicker, faces.
  5. Assembly in code: stitching, subtitles, a permanent, legible "AI visualization" badge, CTA. Render with HyperFrames or Remotion.

Cost

Before steps 1 and 3, estimate the credits and record them in weeks/<W>/costs.md. Regenerate a frame at most 3 times, then ask the human.

Never

  • Present an AI visualization as a completed project (including on Yandex Maps).
  • Generate text, prices or logos inside the image.
  • Insert real people without consent (brand/consents.md).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.