tfcbot/agent-video-team · Archived

director

>- Orchestrate Vidjutsu's tenant-scoped short-form clone workflow: stage a social source, check cloneability, create or reuse a character, compose the first frame, render with Kling 3.0, run QA, and finish post-production.

First seen Apr 12, 2026

Installation

$ npx skills add tfcbot/agent-video-team --skill director

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from tfcbot/agent-video-team · top by installs.

npx skills add tfcbot/agent-video-team

Browse all from tfcbot/agent-video-team

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

License LICENSE
Default branch main
Open issues 0
Status Archived

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,510 B
  • docs SUMMARY.md 235 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 2 installs

SKILL.md

Director

Run the production workflow in dependency order. Reuse every completed artifact and never repeat a paid generation merely because a later stage failed.

Use https://api.vidjutsu.ai with Authorization: Bearer <VIDJUTSUAPIKEY>. The generation and intelligence stages require an active subscription and share fixed daily limits.

Inputs

Collect the TikTok or Instagram source URL, a reusable character ID or character description, optional motion prompt, optional overlay text, and whether external ZapCap captions are wanted.

Pipeline

  1. Stage the source once. Use the platform-specific Vidjutsu download

method and retain its CDN url and assetId.

  1. Check cloneability. Call POST /v1/clones/check with the staged

videoUrl. Treat returned {verdict,score,evidence,model} as authoritative; stop on weak unless the user explicitly overrides it.

  1. Extract frame zero. Call POST /v1/extract with

{ "mediaUrl": sourceUrl, "frames": [0] }; retain frames[0].url.

  1. Resolve identity. Reuse a tenant-owned char_... ID, or call

POST /v1/characters once with {prompt, referenceImageUrl?}.

  1. Create the starting image. Call POST /v1/clones/starting-image with

exactly {firstFrame, characterId, prompt}.

  1. Submit the clone. Call POST /v1/clones/video with

{startingImageUrl, sourceVideoUrl, model:"kling", prompt?}. Source clips must be 3–15 seconds.

  1. Persist and poll the Vidjutsu task ID. Read only

GET /v1/clones/video/{id}. While processing, resume later; on completed, retain videoUrl; on failed, report error.

  1. Run QA. Use /director-qa on the completed Vidjutsu URL.
  2. Post-process. Use /editor-post-production only after QA passes.

Non-negotiable rules

  • Kling 3.0 Motion Control is the sole clone-video model.
  • Never use an alternate model, upstream task ID, or upstream status endpoint

in this workflow.

  • Do not resubmit while a Vidjutsu task remains processing.
  • Vidjutsu Watch and generation operations require an active subscription and

consume daily request capacity rather than per-operation charges.

  • External captions or publishing remain separate and must be labeled as such.