agentsope/SkillAlchemy / skills/agentsope-ai-video-director
agentsope-ai-video-director
UnverifiedDirect and implement programmatic videos from a natural-language sentence: brief, shots, style, timeline, assets, engine-specific coding prompts, draft, review and repair. Use when asked to make a video, product launch, explainer, Reel/Short, kinetic typography or docs-to-video with HyperFrames, Remotion or Motion Canvas, or to improve an existing code-video project. 一句话视频制作与导演工作流。
examples/hyperframes-prompt.md
View on GitHub# Isolated HyperFrames coding-agent prompt
Original authored prompt under C01–C07. Technical bindings are P16/P18/P19 and the current 0.8.142 branch of P20; the older Coleam blank-preview recovery is not applied to this new film. Optional P02 specification detail has unverified visual effectiveness. No creator prompt or asset is copied.
## Prompt to the coding agent
Create a project-local silent evaluation draft from this plan. Target: 15 seconds, 1080×1920, 30 fps, HyperFrames **0.8.142**. Confirm the installed exact version, Node ≥22 and local browser/FFmpeg availability; retain the lock/tree and intended-use rights basis. Do not download a logo, music or external font. The fictitious “AI 产品 / 演示占位” text demonstrates announcement layout and makes no feature claim. Product identity, features and destination link remain unknown.
Create one standalone direct-body composition with matching `data-composition-id="launch-demo"`, `data-width="1080"`, `data-height="1920"`, finite root `data-duration="15"`. The root duration must exist before scripts execute. Use a local authorized GSAP dependency and `gsap.timeline({paused:true})`; finish the whole timeline before assigning `window.__timelines[compositionId] = timeline`. If setup waits for resources, register only at the end. Timed clips have `data-start`/`data-duration` in **seconds**. Keep framework namespaces and readiness flags intact; use application-specific IDs/namespaces, and declare pending application computation through `__hf.buildReady` with unique promise keys. Duration is not resource readiness.
Build the following exact authored visual plan. Background #091018; text #F4F7FC; cyan accent #49DCE8. Title 64 px and supporting labels 32 px at this evaluation size. Use 96 px working inset as a composition token; it is not a platform safe-zone claim. Each shot contains one primary focus and a visible “演示占位” label. S1 [0,3): reveal a rounded outline card, centered portrait close-up, copy “AI 产品”. Fade its inner card from opacity 0 to 1 during the first 0.5 s and hold. S2 [3,7): move the inner card upward within its visual plane and reveal three neutral tiles with “界面示意”; use no metrics or feature names. S3 [7,11): expand an inner panel into the central area and hold “功能信息待确认”. S4 [11,15): hold “了解更多 · 链接待提供” with the card stable, leaving a final 1.5 s readable hold. These timings and style tokens are authored requirements, not distilled creative rules.
Animate inner visual elements. Let the framework own timed-clip visibility and video/audio clock; do not animate timed-clip `visibility`/`autoAlpha`, and do not manually play/seek media. Do not import Remotion frames, Sequence, React hooks or Motion Canvas generator APIs. Seek state must depend on requested time, not wall-clock timers or prior playback. Preserve all facts/unknowns/default labels and the asset ledger in the handoff.
Run these separately from the directory containing the pinned local executable, replacing project/output paths only:
```sh
./node_modules/.bin/hyperframes --version
```
```sh
./node_modules/.bin/hyperframes check <project-dir> --at 0,3,7,11,14.5 --json
```
```sh
./node_modules/.bin/hyperframes snapshot <project-dir> --at 0,3,7,11,14.5 --no-end --describe false
```
```sh
./node_modules/.bin/hyperframes preview <project-dir> --background
```
After eligibility/dependencies are established, strict evaluated export is:
```sh
./node_modules/.bin/hyperframes render <project-dir> --output <output.mp4> --fps 30 --workers 1 --no-best-effort --frames-cache-dir off
```
```sh
ffprobe -v error -show_format -show_streams -of json <output.mp4>
```
The one-worker/strict/cache-disabled choices reproduce an admitted validation profile, not a speed recommendation. Inspect actual file duration, dimensions, fps and audio absence. Inspect visible mounted frames and resource warnings; clean lint does not prove layout or complete assets. Check consecutive frames around 3,7,11 s for unintended flashes/overflow against the authored layout. Compare repeated seeks at the same times. Repair observed layout defects against the supplied shot plan, rerun affected snapshots/probes, and record the change. Do not import a film-local crown, banknote payoff, caption offset, pen treatment or source-note threshold as a new rule.
Deliver source paths, exact engine/lock information, the actual artifact path if one exists, commands/results, QA evidence and limitations. Static/frame samples establish only their coverage. Full-motion pacing requires actual viewing; no audio listening is required only because this particular evaluation draft explicitly requests silence. Report all C07 states separately. If export is unavailable, keep encoded_artifact_available not_executed and explain the concrete dependency gap.
## Binding and handoff references
Read [engine adapters](../references/engine-adapters.md) and [provenance](../references/provenance.json). P16/P18/P19/P20 retain exact documented-contract and local smoke boundaries: research docs at `2dd708e0`, npm gitHead `0b210169`; byte identity covers selected core docs only. The old 2 s silent smoke does not validate this 15 s draft, async assets, full motion or audio. [Manual review](../tests/manual-review.md) describes retained actual output records.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). | Stars: 198k | Updated | Claude CodeCodex | Unverified | Install |
| doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |