> Documentation

Everything you need to know about the agent-media CLI and platform. Install, generate, and manage AI media from your terminal.

# Getting Started

Install the CLI

Install agent-media globally via npm.

npm install -g agent-media-cli

Authenticate

Log in with your account. This opens your browser for a one-time password (OTP) verification.

agent-media login

Your first generation

Generate a video with a single command. The --sync flag waits for the result and returns the download URL.

agent-media generate video -p "robot dancing" --sync

Check your credits

View your remaining credit balance and current plan.

agent-media credits

# Commands Reference

agent-media ships with 28 commands organized into six categories. All commands support --help for detailed usage information.

Account

CommandDescription
loginAuthenticate via browser OTP
logoutClear local session
whoamiShow current user info
creditsDisplay credit balance and usage
planShow current subscription plan
configView or set CLI configuration
profileDisplay user profile details

Jobs

CommandDescription
generateCreate a new generation job
statusCheck status of a job
listList recent generation jobs
inspectShow full job details
downloadDownload generated media
cancelCancel a pending/running job
retryRetry a failed job
deleteDelete a job from history

Models & Info

CommandDescription
modelsList all available models
pricingShow model credit costs

Billing

CommandDescription
subscribeSubscribe to a plan or buy credits (waits for confirmation)
apikeyManage API keys
usageView usage analytics

UGC & Subtitles

CommandDescription
ugcProduce a UGC video from a script
subtitleAdd animated subtitles to any video
persona listList all personas
persona createCreate a persona (voice + face)
persona deleteDelete a persona

System

CommandDescription
doctorRun diagnostic checks
versionPrint CLI version

# Models

agent-media provides access to the following AI generation models. Use the slug when specifying a model in the CLI.

ModelSlugType
AI Video Pro (Kling)video-proVideo
AI Video Cinematic (Veo)video-cinematicVideo
AI Video Realism (Sora)video-realismVideo
AI Video Fast (Seedance)video-fastVideo

Run agent-media models for the latest list and agent-media pricing for credit costs per model.

# Generation

Basic generation

Provide a model slug and a prompt with the -p flag.

agent-media generate <model> -p "your prompt here"

Synchronous mode

Add --sync to wait for the job to complete and receive the download URL directly in your terminal.

agent-media generate video -p "ocean waves at sunset" --sync

Image-to-video

Use --input to provide a reference image for image-to-video generation.

agent-media generate video -p "camera zoom out" --input photo.jpg --sync

Generation options

Fine-tune your generation with additional flags.

FlagDescription
--durationVideo duration in seconds
--resolutionOutput resolution
--aspect-ratioAspect ratio for output
--seedSeed for reproducible results
--inputReference image path
--syncWait for completion

Examples

Video generation

agent-media generate video -p "cinematic drone shot of a forest" --duration 5 --sync

Video with custom duration

agent-media generate video -p "timelapse of city traffic" --duration 10 --sync

Image-to-video

agent-media generate video -p "slow pan right" --input landscape.jpg --sync

# UGC Generation

The UGC pipeline turns a script into a polished video with talking heads, AI B-roll, voiceover, and animated subtitles — all in one command. The pipeline: script → scene splitting → TTS → talking heads + B-roll → assembly → subtitles → output.

Basic UGC generation

Provide a script and the pipeline handles the rest.

agent-media ugc "Ever wonder why some videos go viral? It's not luck..." --sync

AI Script Writer

Skip the script — describe your product and let GPT-4o write it. Costs +5 credits on top of the standard pipeline price.

agent-media ugc -g "A yoga mat with extra grip for hot yoga" --sync
agent-media ugc -g "Sleep tracker app" --product-url https://example.com --sync

With face photo

Add a face photo for talking head scenes. The TTS voice is auto-detected from the face photo when no --voice is specified.

agent-media ugc "Your script here..." --face-url ./photo.png --sync

With persona

Use a saved persona for consistent voice and face across videos.

agent-media ugc "Your script here..." --persona my-brand-voice --sync

Examples

Basic UGC from script

agent-media ugc "Your script text here..." --sync

With face photo for talking heads + auto voice detection

agent-media ugc "script..." --face-url ./photo.png --sync

With custom B-roll images per scene

agent-media ugc "script..." --broll-images "https://img1.jpg,https://img2.jpg" --sync

Dub into Spanish

agent-media ugc "script..." --dub-language es --sync

Full options

agent-media ugc "script..." --face-url ./photo.png --style hormozi --duration 15 --aspect 9:16 --music chill --cta "Follow for more" --sync

UGC options

FlagDescription
-g, --generate-scriptAI writes the script from a one-line prompt (+5 credits)
--product-urlProduct URL for script generation context
--voiceTTS voice (auto-detected from face photo if omitted)
--toneVoice tone: energetic, calm, confident, dramatic
--tts-providerForce TTS provider
--face-urlFace photo for talking heads + auto voice detection
--personaUse saved persona
--styleSubtitle style (hormozi, minimal, bold, karaoke, clean, tiktok, neon)
--durationTarget duration (5/10/15)
--aspectAspect ratio
--musicBackground music genre
--ctaEnd screen text
--modelB-roll model
--broll-imagesCustom B-roll images (comma-separated URLs)
--dub-languageDub final video (BCP-47 code)
--syncWait for completion

Add subtitles to any video

Use the subtitle command to add animated subtitles to an existing video or completed job.

agent-media subtitle <video-or-job-id> --style bold --sync

Available styles: hormozi (yellow karaoke highlight), minimal, bold (neon cyan), karaoke (green pop animation), clean (white on dark box), tiktok (huge white text, thick black stroke), neon (cyan text with magenta karaoke glow).

Persona management

Create and manage personas for consistent voice and face across UGC videos.

Create a persona

agent-media persona create --name "brand-voice" --voice ./sample.mp3 --face ./photo.png

List personas

agent-media persona list

Delete a persona

agent-media persona delete <persona-id>

# Plans & Credits

agent-media uses a credit-based system. Each generation costs a certain number of credits depending on the model and output settings. Choose a plan that fits your workflow.

PlanPriceCredits
Creator$39/mo3,900
Pro$69/mo6,900
Pro Plus$129/mo12,900

Monthly credits reset each billing cycle. Additional credit packs available via the CLI.

# API Keys

API keys allow you to authenticate in scripts, CI/CD pipelines, and automated workflows without an interactive login session. Available on the Pro Plus plan.

List your keys

agent-media apikey list

Create a new key

agent-media apikey create --name "my-key"

Revoke a key

agent-media apikey revoke <key-id>

Use in scripts

Set the AGENT_MEDIA_API_KEY environment variable to authenticate without interactive login.

AGENT_MEDIA_API_KEY=ma_xxx agent-media generate video -p "automated render" --sync

# Gallery & Dashboard

The web dashboard at agent-media.ai/gallery provides a visual interface for managing your generations and account.

Gallery

Browse all your completed generations. Preview videos and images, download files, and delete jobs you no longer need.

Billing

Manage your subscription, view credit balance, transaction history, and access the Stripe Customer Portal.

Settings

Update your profile, manage API keys, and configure account preferences.

# Claude Code Skill

agent-media is available as a Claude Code skill. Install it with one command and your AI coding assistant can generate UGC videos on your behalf.

Install the Skill

Install using the skills CLI (works with Claude Code, Cursor, Copilot, Cline, and more):

npx skills add yuvalsuede/agent-media-skill

Or install manually — download the SKILL.md and add it to your project:

# Download the skill
curl -o SKILL.md https://raw.githubusercontent.com/yuvalsuede/agent-media-skill/main/SKILL.md

# Add to your Claude Code project
mkdir -p .claude/skills
mv SKILL.md .claude/skills/agent-media.md

What you can do

Once the skill is active, talk to Claude naturally:

"Generate a UGC video reviewing Postiz with actor naomi"
"Create a 10-second talking head video about our new feature"
"Show me my credit balance and recent jobs"
"Make a SaaS review video with these screenshots"
"Generate a product demo video with B-roll"

How it works

The skill gives Claude full knowledge of every agent-media command, model, and flag. Claude picks the right actor and model for your request, counts script words to match the duration, checks your credits, runs the generation with --sync, and presents the result — all within your coding session.

# Claude Code Skill

agent-media works as a native skill inside Claude Code. Once installed, Claude can produce complete UGC videos directly from your terminal session.

What Claude does for you

The skill handles the full UGC workflow automatically:

>Picks the right actor and voice for your brand
>Writes scripts that fit the target duration (2.5 words/sec)
>Includes --broll and screenshots for SaaS reviews
>Checks your credit balance before generating
>Runs the command with --sync and returns the video URL

Example session

You: "Make a 10-second UGC review of Postiz with screenshots from their site"

Claude: I'll create a UGC review video for Postiz.
  1. Fetching screenshots from postiz.com...
  2. Picking actor naomi with energetic tone
  3. Writing 25-word script for 10s duration
  4. Running: agent-media ugc "..." --actor naomi --broll --broll-images ... --sync

✓ Video ready: https://...ugc-final.mp4