Video Clipping

Video Clipping is the repurposing primitive for turning long-form video into short-form content. Submit a video URL (a direct file or a YouTube link) and a worker transcribes it, scores the most engaging segments, cuts them into clips, burns in captions, and — by default — reframes them to vertical 9:16. All jobs go through the unified jobs system, and credits are metered per finished clip and input minute at completion (never below twice the job’s true cost).

CLI First

# Auto-clip a long video (async — returns a job id). Vertical 9:16 by default.
naive video clip "https://example.com/podcast.mp4"

# Keep the original 16:9 aspect
naive video clip "https://youtube.com/watch?v=abc123" --aspect 16:9

# Check status of a clip job
naive video clip-status job-uuid-123

Tools

Tool	Type	Description	Cost
`naive_clip_video`	Core	Submit a video for auto-clipping (async)	Metered per input minute
`naive_clip_status`	Core	Check a clip job’s status	Free
`naive_video_compose`	Core	Mux a voiceover track onto a video	Metered on output duration

Creating clips

POST /v1/video/clip submits a video for clipping. The only required field is video_url — a public https URL to a video no longer than 90 minutes (a direct file or a supported host such as YouTube).

curl -X POST https://api.usenaive.ai/v1/video/clip \
  -H "Authorization: Bearer nv_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://example.com/podcast.mp4",
    "aspect_ratio": "9:16",
    "remove_silence": false
  }'

const response = await fetch("https://api.usenaive.ai/v1/video/clip", {
  method: "POST",
  headers: {
    "Authorization": "Bearer nv_sk_your_key",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    video_url: "https://example.com/podcast.mp4",
    aspect_ratio: "9:16",
    remove_silence: false,
  }),
});
const job = await response.json();

Response (202):

{
  "job_id": "job-uuid",
  "status": "queued",
  "message": "Clip job queued"
}

Parameters

Param	Type	Required	Default	Description
`video_url`	string	Yes	—	Public `https` URL to a video (≤ 90 min). Direct file or YouTube.
`aspect_ratio`	string	No	`"9:16"`	Output ratio: `"9:16"` (vertical, face-tracked reframe) or `"16:9"` (keep original)
`remove_silence`	boolean	No	`false`	Trim silences from each clip (adds a small per-clip surcharge)

Only one clip job may be in flight per account at a time. YouTube-source jobs also have a rolling 24-hour bandwidth ceiling to bound proxy cost.

Polling for completion

Clip extraction takes several minutes. Poll the job:

curl https://api.usenaive.ai/v1/video/clip/job-uuid \
  -H "Authorization: Bearer nv_sk_your_key"

const response = await fetch("https://api.usenaive.ai/v1/video/clip/job-uuid", {
  headers: { "Authorization": "Bearer nv_sk_your_key" },
});
const status = await response.json();

Response (completed):

{
  "status": "completed",
  "result": {
    "clips": [
      {
        "title": "The Secret to Scaling",
        "url": "https://.../clip-1.mp4",
        "score": 87
      }
    ]
  },
  "credits_used": 13.2
}

Each finished clip is auto-ingested into your Media Asset Manager with source_type: "video_clipping". Use naive media list --source video_clipping or GET /v1/media?source_type=video_clipping to find them.

Cost

Clipping is metered, not a flat fee. The charge is built server-side from what the job actually produced: 1.6 credits per finished clip plus 0.08 credits per input minute, and never less than twice the job’s own true cost (transcription audio-minutes, scoring tokens, per-clip re-encodes). On top of that sit the two opt-in extras — a flat 4 credits for the 9:16 reframe pass and metered proxy bandwidth on YouTube sources. Credits are charged on completion only; failed or cancelled jobs cost nothing. So a 15-minute upload that yields 5 vertical clips settles at ~13.2 credits (

0.66), and a 90-minute podcast yielding 20 clips at ~43.2 credits (

2.16) — under what Klap/Submagic/OpusClip charge for the same output. Preview the pricing band with GET /v1/video/pricing.

Error handling

Error	Cause	Recovery
`invalid_input`	Missing/invalid `video_url` or bad `aspect_ratio`/`remove_silence`	Provide a public `https` `video_url`; `aspect_ratio` must be `"9:16"` or `"16:9"`
`insufficient_credits`	Not enough credits for the provisional hold	Top up with `naive billing topup`
`rate_limited`	Daily YouTube proxy-bandwidth ceiling reached	Retry after 24h, or use a direct file URL (no proxy)
`provider_error`	Upstream transcription/processing failed	Check the video URL and try again

Typical workflow

Agent task: "Turn this podcast into vertical Shorts"
    |
    +-- POST /v1/video/clip                       --> Submit video_url (9:16 default)
    |   --> job_id: job-uuid, status: queued
    |
    +-- GET /v1/video/clip/job-uuid               --> Poll every 30-60s
    |   --> status: queued -> completed
    |   --> clips[] ranked by score
    |
    +-- GET /v1/media?source_type=video_clipping  --> Auto-ingested to Media Assets
    |
    +-- POST /v1/social/posts                      --> Distribute the top clips

Getting Started

Agents

Agent Profiles

Architecture

Primitives

Video Clipping

CLI First

Tools

Creating clips

Parameters

Polling for completion

Cost

Error handling

Typical workflow

​CLI First

​Tools

​Creating clips

​Parameters

​Polling for completion

​Cost

​Error handling

​Typical workflow

CLI First

Tools

Creating clips

Parameters

Polling for completion

Cost

Error handling

Typical workflow