Media skills

Turn video editing preferences into a reusable skill document. It identifies patterns in pacing, sound, and visual choices, confirms the summary with the user, and saves editing instructions for an agent to follow in future work.

• Documenting a preferred video editing style
• Reusing pacing and audio rules in future edits
• Updating an existing editing style skill

Run an installed OpenStoryline setup to edit local footage through conversation. Start services, submit editing requests, continue revisions in the same session, and check the generated MP4 files and their paths.

• Starting services and editing local footage
• Changing background music through follow-up requests
• Revising script tone in the same editing session

Organize an editing workflow for daily or travel vlogs, including footage review, clip grouping, scripts, voiceover, and music. Build a timeline and render the finished video, choosing optional steps according to the request.

• Editing everyday footage into a vlog
• Arranging travel clips into a coherent story
• Planning an edit when requirements are unclear

Create demo GIFs and MP4 videos from real terminal test runs recorded with tmux and asciinema. Capture HTML viewer screenshots through Playwright, with guidance for selecting visible controls and scrolling the correct page containers.

• Record a real terminal test run as a demo
• Convert a terminal recording into GIF and MP4 files
• Capture HTML viewer entries and diff screens

Generate images and videos through LibTV from text or reference images, and continue existing video sessions. The skill submits requests, tracks generation, retrieves result links, and can automatically deliver finished media to the originating Feishu chat.

• Generating a video from a written description
• Turning a reference image into a video
• Generating or transforming an image

Generate images from text, edit an existing image, or combine up to 14 images into one composition. Choose from three Nano Banana models and specify the output resolution, aspect ratio, and filename.

• Generate an image from a written description
• Change the sky color in an existing image
• Combine multiple images into a collage

Generate images from text, edit an existing picture, or combine up to 14 inputs using Gemini image models. Choose a model, resolution, and aspect ratio, then save the result to a specified output file.

• Generate an image from a written description
• Change colors or details in an existing image
• Combine multiple images into a collage

Turn supplied video material and transcripts into an initial editing plan covering sections, cut points, subtitles, and material sources. The skill clarifies style and delivery preferences, then presents the plan for confirmation.

• Plan an edit from a batch of video material
• Define a preferred video editing style
• Review an already edited video

Check an original video editing plan for timeline consistency, material references, and rights declarations. It outputs continuous frame ranges and keeps unresolved semantic gaps visible, without processing media files or rendering a video.

• Checking a video editing plan before production
• Checking the structure and material references in plan JSON
• Reviewing rights declarations and unresolved gaps

Organize source footage, plan edits, and review delivery for short vertical talking videos. Build the plan from actual transcripts, then produce reviewable local files with approved existing tools, or provide the approved plan and explain missing capabilities.

• Organize footage and usage rights for a vertical talking video
• Plan cuts and captions from an actual transcript
• Check timeline ranges and material references

Audit a supplied PNG for colors outside its palette, translucent edges, canvas dimensions, grid fit, and empty frames. Get a proposed slicing plan without changing the image or accessing the network.

• Check game sprites for colors outside the palette
• Inspect translucent edges around artwork
• Verify sprite sheet dimensions and grid fit

Assemble existing short drama footage into a finished video, with documented cut decisions, subtitles, sound effects, and consistent loudness. Check source files and timing before rendering, then measure the exported video for delivery.

• Assembling generated shots into a finished episode
• Adding script-based subtitles and sound effects
• Adjusting pacing and balancing loudness

Convert approved short drama storyboards and keyframes into copyable video generation prompts. Specify actions, camera movement, dialogue timing, reference images, and start and end states, using the target model’s documented prompt format and duration limits.

• Converting storyboard shots into video generation prompts
• Specifying performances, camera movement, and dialogue timing
• Organizing reference images and missing inputs

Create shot-by-shot plans from Chinese short drama scripts and visual settings. Define timing, framing, movement, and continuity, then write image prompts for each shot’s opening frame, with reference images linked where available.

• Turning a script into a shot list
• Checking positions, eyelines, and prop continuity
• Writing prompts for opening and ending frames

Turn a short drama script into readable visual specifications for characters, outfits, locations, and props. Distinguish reusable identities from changing states, and document visible details that need to remain consistent across scenes and episodes.

• Listing characters, locations, and props from a script
• Deciding whether to reuse an asset or add a variant
• Tracking costume, injury, and prop state changes

Generate images, videos, speech, or music from existing short drama prompts and scripts through configured external services. Preview each job, obtain explicit confirmation, and save the resulting files and execution records in the project.

• Generating images or videos from approved prompts
• Synthesizing dialogue and producing music
• Retrieving submitted jobs after an interruption

Review short drama scripts, visual plans, image and video prompts, and existing media. Identify inconsistencies and missing information, then report evidence, severity, delivery readiness, and the required revisions for the responsible creator.

• Review story logic and dialogue in an episode
• Check continuity across characters, props, and shots
• Review image and video prompts against source documents

Set up and continue short drama or comic-style video projects with organized creative documents, a local progress dashboard, and clear handoffs between stages. Export existing scripts, visual plans, prompts, and finished media for sharing.

• Starting a short drama project
• Checking episode progress and next steps
• Choosing a visual direction

Generate AI-powered podcast-style audio narratives using Azure OpenAI's GPT Realtime Mini model via WebSocket. Use when building text-to-speech features, audio narrative generation, podcast creation from content, or integrating with Azure OpenAI Realtime API for real audio output. Covers full-stack implementation from React frontend to Python FastAPI backend with WebSocket streaming.

The English summary is being prepared. Showing the original description.

操作 chengfeng-videocut 剪辑工作台的已有工程，通过 Runtime CLI 查工程与词句位置、添加或替换画面覆盖层、移除画面、删除明确指定的词句或片段、拆分裁剪移动已有片段、调整画布比例并回读核验。用户说插入片段、删掉这句、移除动画、换个画面、调整顺序或管理时间线时使用；先区分画面覆盖与主轨顺延插入，不负责自动判断口误、创作动画、制作字幕、导出或排查安装故障。

The English summary is being prepared. Showing the original description.

整理、脱敏并上报 chengfeng-videocut 的 GitHub Bug。用户说上报 Bug、反馈剪口播问题、提交 GitHub Issue、这个问题告诉开发者，或要求继续提交已预览的 Bug 草稿时使用。不要用于功能建议、普通排错、代码提交或未获用户确认的自动上报。

The English summary is being prepared. Showing the original description.

VLM evaluation suite for video anomaly detection in smart home camera footage

The English summary is being prepared. Showing the original description.

Rebuild slide images, scanned or image-based PPT/PPTX files, and PDF decks into object-level editable PowerPoint (.pptx), preserving speaker notes when supplied. Use for making visual slides editable or reconstructing slides from screenshots; not for authoring new presentations from scratch.

The English summary is being prepared. Showing the original description.

Use when the user asks to "script this short video", "write a TikTok / Reels / Shorts script", "给这条抖音或视频号视频写脚本", or "fix the hook — viewers drop off in the first seconds"; produces timestamped beat-sheet scripts on the retention-gate model (0-2s hook / 2-5s confirmation / 5-15s payoff / loop-or-CTA) with per-scene script lines, on-screen text for muted viewing, asset keywords, and 9:16 format-param rows per platform norm card for TikTok, Reels, Shorts, 抖音, and 视频号 — plus 2-3 hook options from named hook families and an AI-content disclosure line by default on realistic synthetic media. Spec-only: rendering, TTS, and publishing stay with the user's own tools — no pipelines, no upload automation. Not for creator video briefs — use brief-generator; long-form video SEO belongs to content-writer. 短视频脚本/抖音分镜/开头钩子/竖屏9:16

The English summary is being prepared. Showing the original description.