Writing and planning
Your existing agent and its model connection.
Install the skills.
Connect your tools.
The package provides instructions, scripts, templates, presets, and the local ClipWisp engine. You provide the agent and required programs.
The customer release is in preparation. These instructions describe library version 0.2.1.
Choose your agent.
Your agent must read skill folders, access local files, and run terminal commands. A chat window alone is not enough.
Claude Code, Codex, and other agents use their own settings to find skills and tools. Install into the skills directory that your agent reads.
The installer copies folders. It does not configure your model, enable tools, or install media programs.
Check, then install.
Extract the complete package. Keep each skill with its scripts, references, and templates. The installer requires Python 3.9 or newer.
Run these commands from the extracted package folder. Replace the example path with your agent's actual skill directory.
python3 verify.py python3 install.py --target /path/to/agent/skills python3 install.py --target /path/to/agent/skills --apply
The second command previews the installation. The third copies all 29 entries. The installer stops when it finds a conflicting file.
To keep a skill you already have, add --skip and its name to both installation commands.
Rebuild or reload the agent profile as its host requires. Keep your projects outside the installed skill folders.
Install programs for your selected workflow.
ClipWisp written repurposing
Python 3.10 or newer. Existing transcripts need no video tools.
ClipWisp transcription
Python 3.10+, ffprobe, faster-whisper, and a speech model.
ClipWisp clips
Python 3.10+ and FFmpeg with libx264 and the subtitles filter, including exports without captions.
Carousels and HyperFrames video
Node.js 22+, HyperFrames, FFmpeg, and browser rendering support. Carousels also require Python.
Video editing
Python 3.10+, FFmpeg, and ffprobe. Transcription with video-use needs an ElevenLabs API key. Graphics need HyperFrames.
YouTube thumbnails
Source inspection, frame access, and an image tool with reference support.
Generated music and effects
ElevenLabs access and its required client. Separate provider costs apply.
Research
Web access for current discussions. Hormozi research requires a separate local transcript index.
Provider accounts, API keys, subscriptions, speech models, and render programs are separate from the download.
Start with a transcript.
The ClipWisp text helpers use Python's standard library. Run this check before asking for written content.
python3 skills/clipwisp/scripts/engine/doctor.py --text-only --json
Use ClipWisp Repurpose on this transcript. Create one newsletter and one LinkedIn post. Keep claims tied to the source.
For mixed work, ask creator-content-router to choose the workflow. Use clipwisp for transcription, clips, or repurposing.
Use your channel style.
The YouTube workflow includes light, dark, and podcast examples. A custom profile controls colors, layout, typography, references, and expression style.
Personal presets are optional. Set your own channel profile and use your own identity references.
Use package-youtube-video with my custom channel profile. Recommend a title and thumbnail pair. Wait for approval before generating the image.
If you skipped an older package-youtube-video, use creator-youtube-package. It runs the same workflow.
Current verification
Version 0.2.1 passed file, installation, archive, and text helper checks on macOS on September 24, 2026.
These checks did not render media or configure providers. Linux rendering and complete agent sessions remain untested.
The checked machine lacked FFmpeg subtitle support and faster-whisper. Those workflows require additional setup. Existing transcript workflows passed.
Local tools handle media processing. Your agent and optional services can send source material to external providers.
Try the free YouTube workflow.
The standalone free download is the earlier YouTube skill. It includes source checks, title selection, thumbnail approval, descriptions, and chapters.
The full library adds configurable channel profiles and stronger pose, likeness, and anatomy checks.
Get the free skill →