VividKit Crystal clear AI coding
VI
Choose Your Setup Get Started with AgentKit Continue with ClaudeKit Migrate from ClaudeKit to AgentKit Clean cutover from ClaudeKit What Is AgentKit? How AgentKit Works AgentKit Kits, Skills, Agents & Hooks AgentKit Runtime Support AgentKit Desktop App AgentKit Configuration AgentKit CLI Quick Reference AgentKit Skills Cheatsheet AgentKit Workflows AgentKit Statusline Update AgentKit CLI & Kits AgentKit Helper Clean reinstall of AgentKit AgentKit Troubleshooting What is ClaudeKit? ClaudeKit CLI Guide Command Finder Commands Cheat Sheet Workflow Recipes CLI Quick Reference Claude Mechanics Permission Modes ClaudeKit Hooks Coexistence ClaudeKit x Codex CK → Other coding tools How ClaudeKit Works Inside ClaudeKit Finding Your Unknowns Session Recovery Fix from Logs UI/UX Guide Accounts Switcher IDE Config Remote Control Codex App System Diagramming UI Review Gate Promotions Support VividKit

Quick Ref / Command

Engineer Kit

/ak:ai-multimodal

Multix media analysis and generation

Processes images, audio, video, and documents with the npm-latest @mrgoonie/multix CLI for Gemini vision, OCR, transcription, structured extraction, media optimization, and provider-resolved generation.

01

Resolve CLI

02

Check environment

03

Select modality

04

Resolve models

Rule 01

The npm-latest Multix CLI owns command syntax; this skill owns orchestration and provider setup.

Rule 02

Do not revive a parallel local backend unless an accepted ADR or explicit maintainer decision changes ownership.

Rule 03

Provider catalogs own model IDs, limits, availability, pricing, and deprecations.

Rule 04

Transcript outputs should be Markdown with metadata, chunk status, and timestamped lines.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Resolve CLI Use npx --yes --prefer-online --package=@mrgoonie/multix@latest; do not install or call a global multix binary.
  2. 2 Check environment Require Node.js 20+, provider keys from env/.env/~/.multix/.env, and multix check --verbose before blaming AgentKit.
02

Work

Agent / skill execution

  1. 3 Select modality Choose analyze, transcribe, extract, doc convert, image generation, video generation, MiniMax speech/music, or media optimize based on the artifact.
  2. 4 Resolve models For generation, inspect current help and provider docs for model IDs, pricing, limits, aspect ratios, resolution, duration, and access.
03

Verify

Gate, review, validation

  1. 5 Run command Pass files, prompts, format, output path, and explicit provider/model controls; keep outputs reproducible when needed.
  2. 6 Handle limits When files exceed verified size, duration, or output limits, split media with ffmpeg or a resolved Multix media command and combine segment results.
04

Close

Report, handoff, artifact

  1. 7 Report failures Treat missing keys, FFmpeg, provider access, billing, quota, model support, and npm/offline issues as environment or provider blockers with redacted errors.

Syntax, arguments, subcommands

How to invoke the skill: syntax, positional arguments, shared options, then each subcommand with its own syntax and outcome.

Syntax

/ak:ai-multimodal [file-path] [prompt]

Arguments

[file-path]

Input file path

Optional local image, audio, video, or document to analyze, transcribe, extract from, convert, optimize, or use as context. Name every file that may be uploaded; omit when generating from a text-only brief.

/ak:ai-multimodal ./assets/receipt.png "Extract merchant, date, currency, total, and line items to ./artifacts/receipt.json"
[prompt] Required

Task prompt

Natural-language analysis, transcription, extraction, conversion, optimization, or generation request with output path, format, allowed upload, redaction, rights, and cleanup boundaries. The Skill does not pick a default provider model for generation without live checks.

/ak:ai-multimodal "Generate a 15-second product demo video to ./artifacts/demo.mp4 after resolving a currently available provider model, duration, resolution, aspect ratio, pricing, and retention"

Special Operations

npm-latest contract

Every command resolves @mrgoonie/multix@latest with prefer-online when networking is available.

Provider catalog check

Generation controls come from live provider docs, not from static memory.

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Extract receipt data Recommended 
/ak:ai-multimodal ./assets/receipt.png "Extract merchant, date, currency, total, and line items to ./artifacts/receipt.json; separate unreadable fields from inferred values"
Use when:
Use for OCR, visual understanding, design extraction, or structured extraction from media.
Expected:
Verifies Multix setup, selects the Gemini analyze or extract command family, uploads only the named file, and returns schema-shaped output with uncertainty called out.
Transcribe interview video 
/ak:ai-multimodal ./media/interview.mp4 "Generate a Markdown transcript with metadata, chunk status, and timestamped lines to ./artifacts/interview-transcript.md"
Use when:
Use for audio or video transcription when timestamps and chunk status matter.
Expected:
Runs the Gemini transcribe path through the npm-latest CLI, preserves timestamped Markdown evidence, and reports missing keys, FFmpeg, quota, or provider errors as blockers.
Convert a document 
/ak:ai-multimodal ./docs/report.pdf "Convert this supported document to Markdown at ./artifacts/report.md and include any conversion warnings"
Use when:
Use when the artifact is a document that should become Markdown through Multix doc convert.
Expected:
Checks the npm-latest Multix document command, writes the requested Markdown file, and includes conversion warnings or unsupported-format limits instead of inventing missing content.
Generate provider-backed media 
/ak:ai-multimodal "Generate a 15-second product demo video to ./artifacts/demo.mp4 after resolving a currently available provider model, duration, resolution, aspect ratio, pricing, and retention"
Use when:
Use when image, video, speech, or music generation needs live provider and model capability checks.
Expected:
Inspects current Multix help and provider catalogs before choosing explicit generation controls, records the selected model, and reviews the generated artifact against the brief.

Handled Scope

  • Gemini vision and OCR
  • Design extraction
  • Transcription
  • Document conversion
  • Structured extraction
  • Image/video/speech/music generation
  • Media optimization

Next