菜单
Enhance audio files using Adobe Podcast Enhance (podcast.adobe.com/en/enhance). Uploads an audio file via Chrome browser automation, waits for enhancement, downloads the result, and moves it to the source directory. Use when the user wants to clean up or enhance spoken audio.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Ensure the current project has a docs/coding-standards.md checklist and use it to review code against the default stack (Bun, TypeScript, Drizzle, Biome, Bun-workspaces monorepo, red/green TDD) and quality principles — tidiness, separation of concerns, maintainability, stale assumptions, leftover experimental code, and doc drift. Use this skill whenever the user asks to check coding standards, review code quality, "is this tidy", "does this follow our standards", set up a coding standards doc, scaffold a new project's conventions, or before finishing/merging a change.
Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, or uses any 'grill' trigger phrases.
Figma automation via the figma-use CLI (by dannote). Use when the user needs to create, modify, or inspect Figma designs programmatically. Triggers on requests to render components in Figma, sync design tokens, query or find nodes, export assets, manage variables/pages, compose screens from component instances, delta-sync web app CSS against Figma properties, or troubleshoot Figma CDP connections. Also triggers on mentions of "figma-use", "figma automation", "figma cli", "render to figma", "figma eval", or "design system sync". Never use computer-use (screenshot/click) for Figma — always use figma-use CLI.
Switch to Claude Opus 4.6 model
Fetch articles and save them to a local reading buffer, or compile all unsent articles into a Kindle-optimized PDF digest and email it to Kindle. Use when the user provides URLs to save for reading, or says "send" / "send to kindle" to compile and dispatch the digest.
Launch and control real Chrome Beta with persistent login sessions. Use when automating websites that detect Playwright (LinkedIn, etc.) or when running parallel browser sessions for job applications.
| name | adobe-podcast-enhance |
| description | Enhance audio files using Adobe Podcast Enhance (podcast.adobe.com/en/enhance). Uploads an audio file via Chrome browser automation, waits for enhancement, downloads the result, and moves it to the source directory. Use when the user wants to clean up or enhance spoken audio. |
| compatibility | Requires real-browser (Chrome Beta with remote debugging) |
| allowed-tools | Bash(ls *), Bash(mv *), Bash(ffmpeg *), Bash(pkill *), Bash(sleep *), Bash(agent-browser *) |
| metadata | {"author":"dandaka","version":"2.0","url":"https://podcast.adobe.com/en/enhance"} |
Enhance spoken audio files using Adobe's free AI speech enhancement tool via Chrome browser automation.
Check that the file exists. If the input is a video file (.mov, .mp4, .m4v, etc.), extract audio first (see step 2). Pure audio formats work directly: .mp3, .wav, .aac, .flac, .m4a, .ogg.
Adobe Podcast shows an "Upgrade to enhance videos" modal for video files. Extract audio to .mp3 first:
ffmpeg -i "/path/to/input.mov" -vn -acodec libmp3lame -q:a 2 "/path/to/input.mp3"
Use the same directory and base filename as the original. Keep this intermediate .mp3 until the final video is produced.
Use /real-browser to launch Chrome Beta:
pkill -9 -f "Google Chrome Beta" 2>/dev/null; sleep 2
# then launch in background with --remote-debugging-port=9222 --user-data-dir="$HOME/.chrome-beta-profile"
agent-browser --session main connect 9222
agent-browser --session main open "https://podcast.adobe.com/en/enhance"
Take a screenshot. If the page shows "Sign up / Sign in" instead of the upload area, click the Sign in link and wait for the user to authenticate manually.
The file input has id enhance-file-upload (hidden behind the "Choose files" button). Use agent-browser upload directly on the input — no need for user to manually select:
agent-browser --session main upload "#enhance-file-upload" "/path/to/input.mp3"
If an "Upgrade to enhance videos" modal appears, dismiss it:
agent-browser --session main snapshot -i # find "Dismiss modal" ref
agent-browser --session main click <ref>
Poll with screenshots every 30 seconds. Enhancement shows "Enhancing speech... Separating background..." and takes 30–120 seconds. Done when the audio player appears with Original/Enhanced toggle and "Download" button at the bottom.
Use agent-browser download (not just click) with a full output path:
agent-browser --session main snapshot -i # find "Download" ref
agent-browser --session main download <ref> ~/Downloads/basename-enhanced-v2.mp3
mv ~/Downloads/basename-enhanced-v2.mp3 /original/dir/
If the original input was a video file, replace its audio track with the enhanced mp3:
ffmpeg -i "/path/to/original.mov" -i "/path/to/enhanced-v2.mp3" \
-c:v copy -c:a aac \
-map 0:v:0 -map 1:a:0 -shortest \
"/path/to/original-final.mov"
Output: <basename>-final.mov in the same directory.
The intermediate .mp3 extract can be deleted once the final video is confirmed:
rm "/path/to/input.mp3"
.mov/.mp4 directly triggers a premium upgrade modalagent-browser upload "#enhance-file-upload" <path> — the hidden input accepts setInputFiles from Playwrightagent-browser download <ref> ~/Downloads/filename.mp3Path to the audio or video file to enhance. If not provided, ask the user which file to enhance.