| name | video-kling-master |
| description | Generate a 5-second cinematic video from text using fal.ai Kling 2.1 Master (top tier, better identity preservation than Kling 2.6 Pro). |
| allowed-tools | Bash(generate-video-kling-master:*) |
Text-to-Video — Kling 2.1 Master via fal.ai
Generate a 5-second cinematic video from a text prompt. Kling 2.1 Master is
the highest tier of the Kling line — better identity preservation, sharper
detail, and stronger prompt adherence than the older 2.6 Pro tier.
Usage
generate-video-kling-master "a ghost floating through Mexico City at sunset, cinematic aerial, golden hour"
IMPORTANT: warn the user first
Master tier is slower (~2–4 min). Send a heads-up before calling:
mcp__nanoclaw__send_message({ text: "Generando video tier alto, dame 2-3 min..." })
Then call generate-video-kling-master. Then send the result.
Output & delivery
The script writes an .mp4 to /workspace/group/ and prints the path:
mcp__nanoclaw__send_message({ text: "Listo:", video_path: "/workspace/group/video-kling-master-1234.mp4" })
Specs
- Duration: 5s, 16:9
- Cost: ~$1.40 USD per video ($0.28 per extra second)
- Model:
fal-ai/kling-video/v2.1/master/text-to-video
- Strengths: cinematic motion, identity preservation, prompt adherence
- Weaknesses: text rendering, complex hands, fingers — same as all video models
When to use vs. other video skills
- Want best quality, willing to pay $1.40 → this skill (Kling 2.1 Master, t2v)
- Have a still image + want subtle motion →
video-hailuo (image-to-video)
- Need cheap/fast prototype → use the older
generate-video (Kling 2.6 Pro, ~$0.42)
Prompt tips
- Describe shot composition, lighting, camera movement, mood
- "cinematic aerial", "macro close-up", "tracking shot", "golden hour", "neon"
- Avoid faces and text-heavy scenes
- Keep prompts under ~300 chars — longer prompts dilute attention