Feed Designs You Video Gen

Create vertical algorithm-control shorts: five equal caption beats across the runtime, UI/data-stream montage (WeryAI). Use for attention-economy critique, d...

Security Scan

The skill is plausibly what it claims (a WeryAI-based short-video generator) and only needs the declared WERYAI_API_KEY and node, but there are clear mismatches between the human-facing SKILL.md defaults and the shipped CLI implementation that could cause unexpected behavior.

Purpose & Capability

Name/description, required binary (node), and required env (WERYAI_API_KEY) align with a video-generation skill that calls WeryAI. However, the SKILL.md/default-parameters (duration default = 10, generate_audio default = true) conflict with the included script (buildRequestBody sets duration default to 5 and generate_audio default to false). That mismatch is a substantive inconsistency between the documented behavior and the actual implementation.

Instruction Scope

SKILL.md stays within scope for a video-gen skill: it instructs prompt expansion, timing, assets, and use of the bundled CLI script. It requires only https image URLs and WERYAI_API_KEY. There are no instructions to read unrelated system files or extra env vars. The mandatory prompt-confirmation workflow is explicit.

Install Mechanism

There is no install spec (instruction-only) which is low-risk. The skill does include an executable Node script (scripts/video_gen.js) that will be run; the script has no external package downloads and uses Node 18+ built-in fetch. No remote arbitrary-download install steps are present.

Credentials

Only a single credential is required: WERYAI_API_KEY (also declared as primaryEnv). That matches the documented API usage. No unrelated credentials or file paths are requested.

Persistence & Privilege

The skill is not force-included (always: false) and uses normal autonomous invocation defaults. It does not request to modify other skills or system settings.

What to consider before installing

This skill appears to be a legitimate WeryAI video-generation helper and only needs your WERYAI_API_KEY and Node.js. However, there are mismatches between what the SKILL.md promises and what the included script actually does: the documentation says the default duration is 10s and audio generation defaults to true, but the bundled CLI sets duration to 5s and defaults audio to false if not specified. Before installing or relying on this skill, you should: 1) test with --dry-run or run the script locally to confirm actual defaults and behavior, 2) explicitly pass 'duration', 'model', and 'generate_audio' in requests to avoid surprises, 3) verify you trust the endpoint https://api.weryai.com and use an API key with appropriate, limited privileges, and 4) review content-policy/ethical implications if you plan to generate manipulative or targeted content. These inconsistencies look like sloppy documentation/packaging rather than malicious intent, but they could produce unexpected outputs if you assume the SKILL.md defaults.

You’re being designed (scroll → log → steer)

Template (five equal segments; list exact seconds for the chosen duration, e.g. 10 → 2s each, 15 → 3s each):

• Segment 1 — YOU THINK YOU’RE SCROLLING (or YOU’RE NOT REALLY CHOOSING).
• Segment 2 — THE FEED IS SCROLLING YOU (or equivalent flip).
• Segment 3 — EVERY PAUSE GETS LOGGED.
• Segment 4 — LAYOUT NUDGES YOU.
• Segment 5 — THEY STEER WHAT YOU SEE.

User may swap wording but must keep five windows unless they explicitly compress (then show full custom schedule in confirmation). Default duration is 10 when unspecified.

Dependencies: WERYAI_API_KEY + Node.js 18+. {baseDir}/scripts/video_gen.js + WERYAI_VIDEO_API.md. No other Cursor skills required.

Prerequisites

• WERYAI_API_KEY must be set; Node 18+; https images only.
• Non-empty model.

Security, secrets, and API hosts

• WERYAI_API_KEY: Secret; requires.env / primaryEnv.
• API hosts fixed in video_gen.js.

Prompt expansion (mandatory)

• Duration: Default 10; 5 or 15 when allowed and requested. For five equal beats, each caption span = duration / 5 seconds (state numeric ranges explicitly in the prompt).
• Timeline scaling: If merging to fewer than five lines, list new non-overlapping windows in the confirmation table.

Visuals: Thumb on glass, tiles exploding into data streams, recommendation grid, heatmap vibe, dark mode neon accents, glitch tastefully; hook in the first segment must be visually loud.

Typography: Bold sans, white + black outline, lower third, no overlap between the five lines.

Motion: Continuous—scroll inertia, parallax UI, particles.

Audio: Sub pulse, digital risers if generate_audio true.

Negatives (Kling): illegible HUD text, watermark, crowded captions.

Language: English on screen unless user requests otherwise.

Confirmation: Full expanded prompt with explicit non-overlapping second ranges for all five lines (derive from duration).

Niche checklist

• Reads as critique of systems, not attacks on named individuals.
• Avoid implying illegal surveillance claims unless user explicitly frames fiction.

Workflow

1. Text / image / multi-image (SEEDANCE_2_0, ≤3).
2. Collect optional wording tweaks; default to template lines if user says “use default.”
3. Expand → validate → confirm → wait.
4. [Video](url) only for links.

CLI reference

node {baseDir}/scripts/video_gen.js wait --json '{"model":"KLING_V3_0_PRO","prompt":"…","duration":10,"aspect_ratio":"9:16","generate_audio":true,"negative_prompt":"illegible text, watermark, garbled captions"}'
node {baseDir}/scripts/video_gen.js wait --json '{"model":"SEEDANCE_2_0","prompt":"…","duration":10,"aspect_ratio":"9:16","resolution":"720p","generate_audio":true}'

Full reference: WERYAI_VIDEO_API.md.

Definition of done

Video or error; five timed English lines present in prompt, with numeric second ranges matching the chosen duration.

Boundaries (out of scope)

• Not factual platform internals documentation—rhetorical short only.
• resources/WERYAI_VIDEO_API.md for CLI; no negative_prompt on Seedance.
• Multi-image only SEEDANCE_2_0, max 3.

Example prompts

• Default five-beat feed paranoia, cyberpunk UI
• Softer aesthetic: pastel social app, same five captions
• Screenshot-style image URL → animate with same subtitle schedule
• Parental controls aesthetic but same five-beat ‘who’s really in control’ lines

---

Model and API constraints (frozen for this skill)

Text-to-video

model_key	durations	aspect_ratios	resolutions	audio	negative_prompt	prompt limit
KLING_V3_0_PRO	5, 10, 15	9:16, 1:1, 16:9	(omit)	yes	yes	2000
KLING_V3_0_STA	5, 10, 15	9:16, 1:1, 16:9	(omit)	yes	yes	2000
SEEDANCE_2_0	5, 10, 15	9:16, 1:1, 16:9	480p, 720p	yes	no	2000

Image-to-video

model_key	durations	aspect_ratios	resolutions	audio	negative_prompt	image slots
KLING_V3_0_PRO	5, 10, 15	9:16, 16:9, 1:1	(omit)	yes	yes	single image
KLING_V3_0_STA	5, 10, 15	9:16, 16:9, 1:1	(omit)	yes	yes	single image
SEEDANCE_2_0	5, 10, 15	9:16, 1:1, 16:9	480p, 720p	yes	no	up to 3 images

Recommended models

Tier	model_key	Default duration
⭐ Best	KLING_V3_0_PRO	10
👍 Good	KLING_V3_0_STA	10
⚡ Fast	SEEDANCE_2_0	10 (resolution: 720p)

Default parameters

Field	Value
aspect_ratio	9:16
duration	10
generate_audio	true
resolution	720p (Seedance)
negative_prompt	Kling: illegible text, garbled captions, watermark, HUD clutter

Text-to-video (primary)

Default five-line schedule unless user overrides with a written table.

Image-to-video (optional)

Phone UI still → animate with parallax + same captions.

Tips

• First 2s should sync a hard cut to the first caption.
• Shorter words survive model text rendering better.

Packaged as feed-designs-you-video-gen.