Skills

sfd-api

Installation
SKILL.md

VidJutsu — Video Intelligence API

Watch, extract, transcribe, check. The server-side gaps your agent can't fill.

Authentication

  • Environment variable: VIDJUTSU_API_KEY
  • Header: Authorization: Bearer <VIDJUTSU_API_KEY>
  • If the key is missing, inform the user they must set it.

Base URL

https://api.vidjutsu.ai/v1

Video Intelligence

POST /watch

AI watches the video and responds to a freeform prompt. Returns structured JSON. 10 credits.

Field Type Required Description
mediaUrl string Yes URL of media to watch
prompt string Yes Freeform prompt — what to look for, how to respond

Response (200):

{
  "response": { ... }
}

The shape of response depends on your prompt. Ask for scores, lists, comparisons — AI follows instruction.

POST /extract

Server-side extraction — frames, audio, or metadata. 5 credits.

Field Type Required Description
mediaUrl string Yes URL of media to extract from
frames array or string No Array of integers (e.g. [0, 75, 150]) for specific frame indices, "auto" for evenly-spaced key frames, or "last" for the final frame. Omit for no frames.
audio boolean No Extract audio track
metadata boolean No Extract technical metadata

Response (200):

{
  "frames": [{ "index": 0, "url": "https://cdn.vidjutsu.ai/..." }, { "index": 75, "url": "https://cdn.vidjutsu.ai/..." }],
  "audio": { "url": "https://cdn.vidjutsu.ai/...", "duration": 10.2 },
  "metadata": { "width": 1080, "height": 1920, "fps": 30, "duration": 10.0 }
}

Fields returned depend on what you requested.

POST /transcribe

Speech-to-text with word-level timing. 10 credits.

Field Type Required Description
mediaUrl string Yes URL of media to transcribe

Response (200):

{
  "text": "full transcript here",
  "words": [
    {"word": "full", "start": 0.1, "end": 0.3},
    {"word": "transcript", "start": 0.35, "end": 0.7},
    {"word": "here", "start": 0.75, "end": 0.9}
  ]
}

POST /check

Validate video against spec rules. 5 credits.

Field Type Required Description
spec object Yes VidLang spec JSON to validate
rules array No Array of rule strings to check against (defaults to client's saved rules)

Response (200):

{
  "passed": false,
  "results": [
    {"rule": "MIN_DURATION", "passed": false, "message": "Video is 4.2s, minimum is 5s", "location": "scene[0]", "severity": "error"}
  ]
}

GET /check/rules

Load per-client custom rules. 0 credits.

Response (200):

{
  "rules": [
    {"id": "MIN_DURATION", "params": {"min": 5}},
    {"id": "MAX_DURATION", "params": {"max": 60}},
    {"id": "ASPECT_RATIO", "params": {"ratio": "9:16"}}
  ]
}

PUT /check/rules

Save per-client custom rules. 0 credits.

Field Type Required Description
rules array Yes Array of rule objects with id and params

Upload

POST /upload

Upload media binary. Send file bytes directly with Content-Type header. 0 credits.

Size limits by content type:

Media type Max size Allowed types
Video 32MB mp4, mov, webm
Image 10MB jpeg, png, webp, gif
Audio 25MB mp3, wav, ogg

Videos longer than 5 minutes (300 seconds) will be rejected by /v1/extract and /v1/transcribe.

Response (201):

{"assetId": "asset_...", "url": "https://..."}

POST /upload/url

Upload from an external URL. 0 credits.

Field Type Required Description
sourceUrl string Yes URL of media to upload
contentType string No MIME type (e.g. video/mp4)

Response (201):

{"assetId": "asset_...", "url": "https://..."}

Accounts

Pure data records for organizing content by account. No provisioning or actions. 0 credits.

POST /accounts

Create an account record.

Field Type Required Description
platform string Yes Platform name (e.g. instagram, tiktok)
name string No Display name
niche string No Niche keyword
handle string No Account handle
bio string No Account biography
pfp string No Profile picture URL
linkInBio string No Link in bio URL
tags array No Key-value tags (e.g. [{"key": "tier", "value": "gold"}])

Response (201):

{"id": "acc_...", "platform": "instagram", "name": "fitpage"}

PUT /accounts?id=acc_xxx

Update an account record. Send only the fields to update.

GET /accounts

List accounts. Filter by tag: ?tag.key=value.

Response (200):

{"data": [{"id": "acc_...", "platform": "instagram", "name": "fitpage", "tags": {...}}]}

DELETE /accounts?id=acc_xxx

Soft-delete an account record.

Posts

Pure data records for tracking content. No scheduling or publishing. 0 credits.

POST /posts

Create a post record.

Field Type Required Description
videoId string No VidJutsu video ID
mediaUrl string No Any media URL
caption string No Post caption
brief object No Content brief metadata
tags array No Key-value tags (e.g. [{"key": "tier", "value": "gold"}])
accountId string No Associated account ID

Response (201):

{"id": "post_...", "mediaUrl": "https://...", "caption": "..."}

PUT /posts?id=post_xxx

Update a post record. Send only the fields to update.

GET /posts

List posts. Filter: ?accountId=acc_xxx, ?tag.key=value.

Response (200):

{"data": [{"id": "post_...", "mediaUrl": "...", "caption": "...", "tags": {...}}]}

DELETE /posts?id=post_xxx

Soft-delete a post record.

References

Pure data records for saving reference URLs (inspiration, competitor content, benchmarks). 0 credits.

POST /references

Create a reference record.

Field Type Required Description
url string Yes Reference URL
platform string No Platform name (e.g. instagram, tiktok)
notes string No Free-text notes
tags array No Key-value tags (e.g. [{"key": "niche", "value": "fitness"}])
metadata object No Arbitrary metadata object

Response (201):

{"id": "ref_...", "status": "active"}

GET /references

Get a single reference or list references. Filter: ?id=ref_xxx, ?platform=, ?tag.key=value.

Response (200) — single:

{"id": "ref_...", "url": "https://...", "platform": "instagram", "status": "active"}

Response (200) — list:

{"data": [{"id": "ref_...", "url": "https://...", "platform": "instagram"}]}

PUT /references?id=ref_xxx

Update a reference record. Send only the fields to update.

Field Type Required Description
notes string No Free-text notes
tags array No Key-value tags
metadata object No Arbitrary metadata object
platform string No Platform name

Response (200):

{"referenceId": "ref_...", "updated": true}

DELETE /references?id=ref_xxx

Soft-delete a reference record.

Response (200):

{"id": "ref_...", "status": "deleted"}

Assets

Pure data records for uploaded media. 0 credits.

POST /assets

Create an asset record (metadata only — use /upload for binary upload).

Field Type Required Description
url string No Asset URL
contentType string No MIME type
tags array No Key-value tags
metadata object No Arbitrary metadata object

Response (201):

{"id": "asset_...", "url": "https://..."}

PUT /assets?id=asset_xxx

Update an asset record. Send only the fields to update.

GET /assets

List assets. Filter: ?tag.key=value.

GET /assets?id=asset_xxx

Get a single asset.

DELETE /assets?id=asset_xxx

Soft-delete an asset.

Billing

POST /credits

Create an API key via Stripe checkout.

Field Type Required Description
email string No Email for key recovery
credits number No Initial credits (default 100)

Response (200): Stripe checkout session URL.

POST /subscribe

Create a Stripe checkout session for the $99/mo subscription. 0 credits.

Field Type Required Description
email string Yes Email for the subscription

Response (200): Stripe checkout session URL.

GET /credits/status?session=SESSION_ID

Check Stripe session status and retrieve API key after payment.

GET /balance

Credit balance and client ID.

Common Workflows

QA gate before posting

1. POST /watch {mediaUrl: "https://...video.mp4", prompt: "Score quality 1-10. List issues with severity for: face consistency, artifacts, motion, audio sync."}
2. If score >= 8, proceed to post

Deep analysis for content strategy

1. POST /watch {mediaUrl: "https://...", prompt: "Analyze this video. Return: hook text, format, pacing, transitions, CTA, tags."}
2. Use result to inform next video

Speech verification

1. POST /transcribe {mediaUrl: "https://...video.mp4"}
2. Compare transcript word-by-word against intended dialogue

Frame chaining for multi-scene videos

1. POST /extract {mediaUrl: "https://...scene1.mp4", frames: "last"}
2. Use returned frame URL as starting image for next scene

Spec validation

1. GET /check/rules — load your saved rules
2. POST /check {spec: { ... vidlang spec JSON ... }} — validate against rules
3. Fix failing results, regenerate if needed

Track content with accounts and posts

1. POST /accounts {platform: "instagram", name: "fitpage", handle: "@fitpage", tags: [{"key": "tier", "value": "gold"}]}
2. POST /posts {mediaUrl: "https://...", caption: "...", accountId: "acc_...", tags: [{"key": "batch", "value": "april-2026"}]}
3. GET /posts?accountId=acc_xxx
Related skills

More from tfcbot/shortform-distribution-skills

Installs
3
Repository
tfcbot/shortfor…n-skills
First Seen
Apr 2, 2026
Security Audits
Gen Agent Trust HubPass
SocketPass
SnykWarn