← Back to list
video-script Checked
A skill from the GitHub repository zenstory-ai/video-recap-skills.
8.8Overall rating
Installation
🤖 Install via AI
Copy the prompt below and send it to your AI assistant (e.g. Claude Code) — it will follow the instructions and install automatically.
Install the "video-script" skill by following the instructions at https://skill123.me/install/video-script.
⌨️ Command line install
Run in your terminal — downloads and installs to ~/.claude/skills/.
curl -fsSL https://skill123.me/install/video-script.sh | bash
📦 Download ZIP
Download the zip and extract it into your skills directory (e.g. ~/.claude/skills/), then restart your session.
⬇ Download v1.0.0 · 86 KBAbout this skill
Overview
A skill from the GitHub repository zenstory-ai/video-recap-skills.
Source
- Repo: https://github.com/zenstory-ai/video-recap-skills
- Path: skills/video-script
Score breakdown
Trigger
7.0
Precise stage description with work_dir preconditions and bilingual trigger words; no explicit When-Not in the description.
Structure
9.3
248-line SKILL.md over 17 scripts and 2 purposeful references; clean separation, entry points have shebangs.
Workflow
9.2
Thoughtful creative workflow: five sequential roles, narration_job taxonomy preventing default wall-to-wall narration, counterfactual self-review checklist, two-tier verification (advisory LLM review with override audit log + deterministic validate.py aligning narration to silence windows), and revision freeze semantics.
Content
9.0
Concrete JSON schemas and reason-string conventions; consistent cross-file terminology with the sibling skills.
Engineering
9.0
Valid frontmatter, shebangs on entry points, referenced files exist, MiMo env surface documented in lib.py comments.
Security
9.5
Capability manifest matches declared scope: single declared cloud destination (MiMo) used only for advisory review, fail-open by default, API errors sanitized before logging, keys env-only; one point off because the cloud review call is not visible from the description alone.
Creative planning + narration-writing stage of the zenstory-ai video-recap pipeline: director/story-editor roles, audio_owner and narration_job taxonomies, timestamped narration.json with advisory LLM review plus deterministic validation. Capability manifest (6.0): reads only work_dir understanding artifacts (brief, VLM/ASR indexes, run manifest) and env API keys; side effects confined to work_dir (story plan, visual/audio board, style card, clip plan, narration, review and lint outputs); network destinations: api.xiaomimimo.com / token-plan clusters only, for the advisory semantic review (sends narration text plus VLM/ASR evidence excerpts), declared in SKILL.md section 6.1 with fail-open default and an auditable override log; credentials env-only with sanitized API errors; no persistence. Capabilities match declared scope in SKILL.md; description-level gap: the description's 'review' does not reveal the cloud LLM call.