Video Generator CLI

Guide for running the Video Generator CLI commands.

Security Scan

The instructions and declared runtime requirements in SKILL.md are mostly reasonable for a video-generation CLI, but they conflict with the skill registry metadata and leave ambiguous where voice synthesis happens — the mismatch and ambiguity warrant caution before use.

Purpose & Capability

SKILL.md requires node, npm, ffmpeg, python and a PEXELS_API_KEY which all make sense for a video generator that downloads stock footage and assembles + transcodes video. However the registry-level requirements reported above list no required binaries and no required env vars. That mismatch (manifest vs SKILL.md) is incoherent and should be resolved — a consumer cannot tell which requirements are authoritative.

Instruction Scope

The instructions stay within the scope of running a local CLI: wiping a .video-cache.json, cleaning public/ folders, reading input/input-scripts.json, downloading stock footage, rendering segments, and stitching with ffmpeg. One ambiguity: SKILL.md says it 'generates voiceovers' and names AI voice IDs but does not declare what TTS service or credentials (if any) are used. That could require additional secrets or network calls not declared in the file.

Install Mechanism

This is an instruction-only skill with no install spec and no code files. That is low-risk from an install/extraction perspective — nothing is downloaded or written by a packaged installer described here.

Credentials

SKILL.md lists PEXELS_API_KEY as a required env var which is proportionate for downloading stock footage. But the registry metadata reported earlier lists no required env vars. The mismatch is concerning because it hides a declared credential requirement from the registry-level summary. Also the voice synthesis step mentions AI voice IDs but no TTS credentials are declared; additional credentials may be required but are not specified.

Persistence & Privilege

The skill is not always-enabled, is user-invocable, and has no install script that writes persistent configuration. It does not request elevated persistence or system-wide configuration changes.

What to consider before installing

Before using this skill: (1) Confirm which manifest is authoritative — the registry summary or the SKILL.md — because SKILL.md requires node/npm/ffmpeg/python and PEXELS_API_KEY while the registry metadata shows none. (2) Inspect the project's package.json and the actual CLI scripts (npm scripts) referenced (generate, resume, segment, dev) in a trusted code review environment before running them; the SKILL.md explains behaviors such as wiping .video-cache.json and clearing public/videos and public/audio, so back up any important files beforehand. (3) Ask or inspect where 'voiceover generation' is performed — it may call an external TTS service and require additional credentials you should not supply blindly. (4) Limit the PEXELS_API_KEY scope (use a dedicated key) and verify network endpoints used by the tool. (5) Because this skill is instruction-only (no bundled code), prefer installing and reviewing the actual project source from a trustworthy repository prior to execution. If you cannot resolve the manifest mismatch or determine the TTS behavior, treat this skill with caution and avoid running destructive commands (generate/build) until you can audit the code.

Video Generator CLI Operations Skill

This skill explains the different terminal commands available to run, resume, or fix video generations.

1. Standard Generation Commands

These commands are used for a "Fresh Start." They clean up old temporary files to prevent conflicts.

npm run generate (or npm run build)

• What it does:
  1. Wipes the .video-cache.json file.
  2. Cleans the public/videos and public/audio folders.
  3. Parses your script from input/input-scripts.json.
  4. Downloads new stock footage and generates voiceovers.
  5. Renders the video scene-by-scene (Segmented Mode).
• When to use: When you have a new script or want to see fresh visuals for an existing script.

---

2. Advanced Recovery Commands

These commands save time by reusing already-completed work.

npm run resume

• What it does:
  1. Preserves everything in public/ and the .video-cache.json.
  2. Only downloads visuals/audio that are missing.
  3. Only renders segments that don't exist yet in the output folder.
  4. Automatically stitches everything together at the end.
• When to use: If your computer crashed, or you lost internet during a long generation. It will pick up exactly where it left off.

npm run segment

• What it does:
  1. Skips ALL asset fetching and generation.
  2. Jumps directly to "Phase 7: Assembly."
  3. Takes the existing .mp4 segments in your output folder and merges them.
• When to use: Use this if the video was rendered successfully but the final "stitching" failed or if you want to manually re-assemble the segments.

---

3. Developer / Debugging Tools

npm run dev

• What it does: Starts the local asset server.
• When to use: Only needed if you are manually debugging the Remotion components in the browser.

---

4. Operational Summary

Command	Action	Fresh/Resume
npm run generate	Full Process	Fresh (Wipes old data)
npm run build	Full Process	Fresh (Alias for generate)
npm run resume	Skip Existing	Resume (Saves time)
npm run segment	Assembly Only	Fix (Stitching only)

---

5. Configuration (JSON Script)

When editing input/input-scripts.json, agents can use these newer properties:

Property	Type	Description
showText	Boolean	Set to false to disable on-screen subtitles.
defaultVideo	String	Filename in input/input-assests/ to use as a visual fallback.
orientation	String	portrait or landscape.
voice	String	AI voice ID (e.g., en-US-JennyNeural).