Video generation
Marketing skills for AI agents — paid ads, social media, SEO, competitor research, creative generation, email, analytics, and more. Powered by Hyper MCP.
npx -y skills add hyperfx-ai/marketing-skills --skill video-generationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
End-to-end AI video production through the Hyper MCP — text-to-video and image-to-video generation (Sora, Veo, Seedance), scene chaining, video analysis, transcription, subtitles, TikTok / karaoke captions, voiceover (TTS), audio mixing, clipping, stitching, and text overlays. Use when the user asks to generate a video, create UGC, scene-chain, add captions or subtitles, add narration, stitch clips, clip a podcast highlight, or do any AI video editing.
SKILL.md
5.7 KB, as published. Nobody here has run it
Video Generation & Editing
Guide for generating, editing, analyzing, and post-processing videos using AI models and FFmpeg-backed tools exposed through the Hyper MCP.
Requirements
This skill assumes the Hyper MCP is connected to your agent so the tools below are available. The underlying providers (OpenAI Sora, Google Veo, ByteDance Seedance, OpenAI TTS, transcription, etc.) are configured under your Hyper integrations.
Tool surface
| Group | Tools |
|---|---|
| Generation | videos_generate, sora_videos_remix, sora_videos_delete |
| Analysis | videos_analyze, videos_frames_capture, videos_transcribe |
| Subtitles & captions | videos_subtitles_generate, videos_subtitles_burn, videos_captions_burn_highlighted |
| Audio | audio_speech_generate, videos_audio_add |
| Editing | videos_clips_extract, videos_stitch, videos_text_overlays_add |
Out of scope
- Image generation, ad creative composition, brand extraction — use
image-generationorad-creative-generation. - Posting finished videos to social platforms — use
tiktok,instagram, orlinkedin. - Running paid video campaigns — use
google-ads,meta-ads,tiktok-ads.
Available Tools
| Tool | Purpose | Runs in Background |
|---|---|---|
videos_generate | Generate video from text / image prompt | Yes |
sora_videos_remix | Modify existing Sora video | Yes |
sora_videos_delete | Delete a Sora video | No |
videos_frames_capture | Extract frame as image | No |
videos_analyze | Watch and understand video content | No |
videos_transcribe | Extract audio transcript | No |
videos_subtitles_generate | Create SRT / VTT subtitle file | No |
videos_subtitles_burn | Burn subtitles onto video | Yes |
videos_captions_burn_highlighted | TikTok / karaoke-style word-by-word captions | Yes |
audio_speech_generate | Generate voiceover audio from text | No |
videos_audio_add | Add / replace audio track on video | Yes |
videos_clips_extract | Extract a time segment from video | Yes |
videos_stitch | Concatenate multiple clips | Yes |
videos_text_overlays_add | Add text / titles to video | Yes |
Video Understanding
You can watch and analyze any video using videos_analyze. This sends the video to a multimodal AI that sees both visual and audio content.
When to use videos_analyze
- After generating a video: check if it matches your intent
- Before stitching: verify scene consistency across clips
- Quality review: check for glitches, character drift, lighting issues
- Content understanding: "what happens in this video?"
Analysis Types
videos_analyze(file_id="...", analysis_type="general")
videos_analyze(file_id="...", analysis_type="quality_review")
videos_analyze(file_id="...", analysis_type="scene_breakdown")
videos_analyze(file_id="...", question="Does this match: [original prompt]?")
Self-Review Workflow
Always review generated videos before delivering to the user:
result = videos_generate(prompt="...", model="veo-3.1-generate-preview")
review = videos_analyze(file_id="video_file_id", analysis_type="quality_review")
# If issues found, regenerate with adjustments. If quality is good, proceed to editing.
Routing table
All reference files live in
references/. Read them atreferences/<file>(e.g.references/generation.md).
| The user wants to… | Read these files first |
|---|---|
| Generate a video (any model) | references/generation.md — model selection, parameter matrix, prompt templates |
| Build a longer multi-scene video | references/generation.md — script planning + scene chaining |
| Add subtitles / captions / voiceover / overlays, or clip a video | references/post-production.md |
| Produce UGC / TikTok content end-to-end | references/ugc-video.md (ugc_videos_create modes) → references/workflows.md |
| Shape a prompt for a specific model (Sora / Veo / Seedance / Kling) | references/video-prompting.md |
| Turn a podcast / long video into short clips | references/workflows.md → references/post-production.md |
| Understand or QA an existing video | Use videos_analyze (see Video Understanding above) |
Best Practices
- Review before delivering: always use
videos_analyzeto check your output. - Maintain visual consistency: use the same character descriptions, lighting, and style across all scenes.
- Plan transitions: design the end of each scene to flow into the next.
- Batch similar scenes: generate scenes with similar settings together.
- Review before chaining: check each scene before using its last frame for the next.
- Use single-variable iteration: remix / regenerate by changing one variable at a time.
- Add captions for accessibility: use the subtitle pipeline for all UGC content.
Gives 0 of the 12 instructions most video audio skills give
Counted across 621 of the 795 authors here whose files we hold, read 2026-08-06
- read individual rule files for detailed explanationsin 21 of 621, across 9 files
- Use WAV PCM 16kHz mono audio formatin 13 of 621, across 4 files
- render final videoin 13 of 621, across 6 files
- use this skill when dealing with Remotion codein 11 of 621, across 4 files
- save generated audio to a WAV filein 11 of 621, across 4 files
- handle conversion errors gracefullyin 10 of 621, across 6 files
- add captions to videos alwaysin 10 of 621, across 4 files
- generate music from text descriptions using MusicGenin 9 of 621, across 2 files
- do not skip pipeline layersin 9 of 621, across 3 files
- do not make one tool do everythingin 9 of 621, across 3 files
- never ask the user to paste their full API keyin 9 of 621, across 3 files
- use azure document intelligence for complex pdfsin 9 of 621, across 4 files
Said here and by no other author read
- read the routing table references before executing tasks
- review generated videos before delivering
- use videos_analyze to check output quality
- maintain visual consistency across scenes
- plan transitions between scenes
- generate scenes with similar settings together
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.