Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add MendesCorporation/ai-promo-video --skill create-ai-promo-videogit clone --depth 1 https://github.com/MendesCorporation/ai-promo-videoWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mendescorporation/ai-promo-video/create-ai-promo-video)<a href="https://agentmods.dev/skills/mendescorporation/ai-promo-video/create-ai-promo-video"><img src="https://agentmods.dev/badge/skills/mendescorporation/ai-promo-video/create-ai-promo-video/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mendescorporation/ai-promo-video/create-ai-promo-video"><img src="https://agentmods.dev/badge/skills/mendescorporation/ai-promo-video/create-ai-promo-video.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00116 | $0.02941 |
| Opus 5 | $0.00058 | $0.01470 |
| Sonnet 5 | $0.00023 | $0.00588 |
| Haiku 4.5 | $0.00012 | $0.00294 |
Grade A, and why
create-ai-promo-video scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Create AI Promo Video
Treat the work as motion direction and compositing, not as a screen recording or slide deck. The host model is the creative director and writes the unique scene source; local tools execute deterministic instructions.
Composition Model
Every new production is authored directly as one unrestricted Revideo TypeScript composition. For a short promo, keep one master engine scene and organize its narrative shots as modular generators or groups so persistent camera, environment, light, particles, shared objects, and transition overlaps can cross logical shot boundaries. Scaffold once, search the reusable motion vocabulary, add only the source helper groups selected for this production, and use only components that serve the specific story. Every library rig is optional: customize, combine, replace, or ignore it, and write any compatible nodes, timing, geometry, transitions, camera systems, shaders, SVG, or Three.js behavior the scene needs. Read component-library.md and advanced-motion.md.
Library-first does not mean template-first: never render a generic draft and replace its contents later. The neutral scaffold contains no visual design. Decide the art direction before choosing components, and do not inherit sample palettes, copy, card counts, layouts, music, or scene arcs from prior videos.
The version 1 JSON renderer is legacy compatibility only. Use it when the user supplies an existing v1 spec or explicitly requests maintenance of that format. Never use it to start a new production or as a disposable intermediate render. Read video-spec.md only for that compatibility case.
Workflow
- Inspect the product, positioning, brand, and strongest proof. Define one audience, tension, promise, proof, and an appropriate CTA or resolution using story-direction.md.
- Define a distinct art direction, delivery format, and timed beat plan. Choose landscape, portrait, or square composition intentionally; for social vertical work, choose the platform profile and verify current safe areas. Choose story functions and shot scales for this brief instead of copying a fixed hero/product/cards/CTA sequence. Record the intended energy curve, camera logic, transition motive, and—when prior work or references are available—what makes this production visually different. Before spreading a metaphor across the film, state why it belongs to this audience, promise, product, and brand. Read vertical-and-captions.md for portrait composition or speech-led captions.
- When real product proof is part of the story, create a private capture spec. Keep credentials in gitignored
*.local.json. Capture screenshots or meaningful live flows as described in capture-spec.md. Skip capture when the authorized brief is intentionally abstract, brand-led, or entirely based on supplied media. - When recording product interaction, prefer a clean source with the browser pointer hidden. Add a designed cursor, trace, click pulse, focus ring, or camera response only when the interaction needs it; none is mandatory decoration.
- When outside media strengthens the story, search user-approved local folders and the aggregated licensed providers. For stock video and photography, prefer Pexels and Pixabay when their free API keys are available; use Openverse and Wikimedia Commons for open-license breadth, SVGs, illustrations, animations, and fallback footage. Review previews and source pages, reject unsafe licenses, and download selections with their attribution manifests. Read free-media-sourcing.md.
- Write musical intent in plain language before searching: emotion, energy curve, instrumentation to prefer or avoid, pacing, and edit accents. For online music, search the Freesound catalog first through Openverse's
freesoundsource, then compare Jamendo and Wikimedia Audio; include user-approved local folders when available. Do not call the Freesound API directly in the default commercial workflow unless the user has separately authorized its API terms. Bundled CC0 tracks remain available only when explicitly included. Compare at least three viable candidates, analyze the finalists, verify the license, and choose by fit rather than result order. No track is a default or fallback. Read music-sourcing.md. - If speech captions are requested, call
prepare_caption_timingbefore authoring them. Prefer exact word timestamps; when only cue timing exists, keep the returnedcue-interpolatedprecision label and do not claim exact alignment. Callscaffold_advanced_videoonce with the intended format and platform to create the minimal neutral Revideo composition. Complete the generatedmotion-plan.jsonbefore authoring: visual concept, movement principle, every shot and layer's motion, focal regions, velocity bridges, exact transition/settle moments, and every intentional still interval with a reason. Every non-final logical shot must declareboundaryToNextascontinuous,motivated-cut, orintentional-stop, including the continuity carrier when applicable and a concise intent; continuity does not mean constant motion. Callvalidate_motion_planand repair every exact error before the first render. Querysearch_motion_componentsfor the shot's narrative needs and inspect selected entries withget_motion_component. Then calladd_advanced_video_helpersonce with only the exact source groups those selected mechanisms require; dependencies are resolved automatically, authored files are never overwritten, and unused helper files must not be added. When a tool, component, transition, or fragile parameter is unfamiliar or uncertain, callhelpfor that exact target before authoring it—for examplecomponent:continuous-camera-path,topic:layout-audit, ortool:render_advanced_video. Search help only when the exact id is unknown; do not load the complete help catalog. Author the actual TypeScript timeline from selected primitives plus custom code in the same files. Register every critical text, logo, CTA, caption, and focal product ref withcreateReviewRegistry, including its source label, constraint, permitted overlaps, and centering intent; keepReviewOverlaylast in the scene tree. Vary compositions, camera behavior, transitions, typography systems, particle seeds, and palettes according to the art direction. Do not create or render a template that will be replaced later. - Call
render_advanced_video, probe the result, then callcreate_visual_review_packwithprojectFile,motionPlanPath, the same render variables, and every planned scene boundary. InspectsourceQuality; then pass every returned exactevidenceFrames,settledFrames, overview sheet, and transition sheet toread_visual_files. Classify each marked overflow, collision, centering delta, or focal-motion lull as intentional or material; do not infer quality from successful rendering, metadata, or one contact sheet. Apply the quality gate in motion-quality.md. - Record each material anomaly. Before revising existing work, call
list_advanced_video_filesandread_advanced_video_file, then patch exact source, render only the affected interval, and create a new visual review pack for that interval. Replace the exact range only after the revised pack passes. Read revision-workflow.md. - Track every generated review, audit, preview, cache, and intermediate-render path during production. After the accepted final passes one last probe, call
clean_delivery_outputwith the exact project root, the actual output directory, requested final filenames, and any nonstandard temporary paths. Project and output may be separate: an external host-leveloutputsdirectory defaults toshared, which verifies the final but never removes sibling outputs; useownedonly for a dedicated per-production delivery directory. Never relocate a valid final merely to satisfy cleanup. The cleaner recursively removes generated directories at any project depth while protected source/media directories remain untouched. Keep the editable composition plus required render inputs, capture masters, downloaded media, license sidecars, attribution manifests, music licenses, references, and configuration. Never delete user-provided or irreplaceable inputs unless explicitly authorized.
What ships with it
11 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 616 B
- references/advanced-motion.md 8.9 KB
- references/capture-spec.md 1.6 KB
- references/component-library.md 5.4 KB
- references/free-media-sourcing.md 3.6 KB
- references/motion-quality.md 5.9 KB
- references/music-sourcing.md 3.8 KB
- references/revision-workflow.md 1.2 KB
- references/story-direction.md 1.6 KB
- references/vertical-and-captions.md 3.4 KB
- references/video-spec.md 1.3 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 60 lines · 116 tokens per session scan A 29cd920e6ef1
create-ai-promo-video is a skill published in the GitHub repository MendesCorporation/ai-promo-video (15 stars, last pushed 1mo ago), licensed MIT. It adds 116 tokens to every session and 2,941 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
chat-pet-sprite-creation
Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…