Agent skill · creative production · jeremylongshore

klingai-batch-processing

Process multiple video generation requests efficiently with Kling AI.

Why this skill is useful

Adds executable scripts for batch processing and API interaction with Kling AI that aren't in public docs.

What it needs

Requires npm installed locally. Requires kling-ai account access. About 4k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.

What this skill does

Kling AI Batch Processing Overview Generate multiple videos efficiently using controlled parallelism, rate-limit-aware submission, progress tracking, and result collection. All requests go through https://api.klingai.com/v1. Batch Submission with Rate Limiting Collect Results Async Batch with asyncio Batch with Callbacks (No Polling) Cost Estimation Before Batch Prerequisites An approved batch manifest with a unique batch ID, a bounded count, model, duration, mode, destination, and credit ceiling. Prompts and reference media must be synthetic or rights-cleared, and the request must pass the provider's content policy review. Do not submit real people's likenesses, private data, or copyrighted material without documented permission. Use a sandbox project and draft/watermarked outputs for the first canary. Store KLINGACCESSKEY and KLINGSECRETKEY in the approved secret manager; never place them in prompts, source control, or logs. Instructions 1. Validate the manifest before any request: reject missing rights/consent, disallowed content, unapproved destinations, duplicate batch IDs, and a projected credit total above the approved ceiling. 2. Run one synthetic canary with the lowest-cost permitted mode. Confirm the model, duration, aspect ratio, callback destination, watermark/draft status, and contactsexported=0-style no-export invariant before expanding the batch. 3. Submit only the approved count with bounded concurrency and pacing. Record opaque task IDs and an idempotency key; never log prompts, source media, callback secrets, or result URLs. 4. Poll or receive callbacks with a timeout and a retry budget. Hold successful outputs in quarantine until an owner reviews content policy, rights, quality, and cost results. 5. Promote approved outputs to the allowlisted destination, then expire temporary artifacts and access. Keep only a redacted receipt and the rollback reference. …

How to use it

Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:

@skills jeremylongshore/klingai-batch-processing--af1942

View the source on GitHub

Browse the @skills marketplace