Skip to main content
Glama
Poppify

Poppify Studio

Official
by Poppify

Poppify — AI Reels Maker & Video Generator (MCP Server + Claude Code Plugin)

Make reels worth sharing, from inside Claude. The AI video generator and reels maker for Instagram, TikTok, YouTube Shorts and Facebook. Give it 1–10 photos or just a topic and get back a captioned 9:16 reel — Live Motion that animates the subject inside a still, AI-generated scenes you don't already have, cinematic camera moves, AI voiceover and music, assembled into one edit and published to your channels. $0.06 base render. 50 free seeds on signup, up to 150 once you connect a social account. No subscription.

What you can make

  • Live Motion — image-to-video animates the subject inside a still (breath, blink, a turn of the head) while the camera move layers on top. ~10 seeds; cache hits are free.

  • Images & shots — generate the scenes you don't already have, or build the reel around your own photos.

  • Music & voiceover — a soundtrack and AI narration, mixed with ducking and start-cue alignment.

  • The edit — frames, captions, music and voice assembled into one reel. This is the step that makes a reel rather than a pile of assets.

  • Share on social — publish straight to Instagram, TikTok, YouTube Shorts or Facebook. Publishing is free.

Poppify is an AI video generator built for reels specifically. It composes — FFmpeg motion, library-first asset matching, recipe-driven narrative and on-screen text — rather than hallucinating every frame, which is why a finished reel is ~$0.06 instead of dollars-per-second — the full cost breakdown of AI video generation shows where the difference comes from. Text-in works too: start from a topic with no photos at all and it generates the scenes.

Where it fits in a Claude Code stack: pairs with Postiz for cross-platform scheduling and Windsor.ai for attribution. Use HyperFrames when you want to code video in HTML.

Not for: avatar-based presenter video (use HeyGen / Synthesia), 4K horizontal cinema, or sub-4-second clips.

Built for: anyone who posts every day and wants the reel finished rather than started.

Guides

Written for the formats people actually ask for:

What's inside

Component

Purpose

MCP server (https://poppify.ai/mcp, HTTP)

All Poppify tools: register, start_session_from_photos, apply_session_patch, update_slides, add_slide_image, confirm, get_result, publish_post (post now / schedule to connected channels, linked accounts), portfolio (list / switch), etc.

/poppify:make-reel slash command

Guided session flow — photo-led or topic-led entry, free customization loop, paid confirm.

/poppify:troubleshoot slash command

Symptom triage when a render came out wrong.

/poppify:verify-render slash command

Optional (shell + ffmpeg): download + ffprobe + frame-extract verdict on a finished MP4.

poppify-build-reel skill

The canonical photo-led / topic-led flow — when to use which tool, where the free vs paid boundaries are.

poppify-text-card skill

Render a pixel-perfect text card (terminal frame, code, stat callout, headline) locally via HTML/CSS + a headless browser, cross-platform. Shell-capable clients only; shell-less clients let the composer draw the caption.

poppify-live-motion skill

Veo live-motion prompting playbook — plain subject animation AND first/last-frame transitions (composition-locked end frames, journey overridePrompt, failure modes).

poppify-render-debug skill

Optional (shell + ffmpeg): download the finished MP4, run ffprobe, extract frames, surface a verdict.

poppify-troubleshoot skill

Decision tree: symptom → root cause → action for missing audio, wrong colors, dropped captions, stuck renders.

poppify-schema-introspect skill

How to verify the deployed MCP schema when a parameter looks silently dropped.

poppify-schedule-optimizer skill

Engagement-optimized publishing — recommendedSlots (real engagement history vs platform norms), scheduledAt:"best", slot assessment, multi-post spacing.

Related MCP server: AITuber MCP Server

Install

In Claude Code:

/plugin marketplace add Poppify/poppify-claude-plugin
/plugin install poppify@poppify

That's it. After install:

  1. register() (optionally { label: "claude" }) mints a wallet and returns an apiKey + a signupBonusUrl (50 free seeds, up to 150 after connecting a social account — claim before paying).

  2. Use the poppify-build-reel skill (Claude will auto-invoke when you ask it to "make a reel") to drive the rest.

What does it cost?

  • MCP install: free

  • All non-generation tools (search library, apply session patches, attach audio, update slides, register, wallet balance / top-up / link, portfolio list/switch, publish + schedule): free

  • confirm render: 1 seed base

  • add_slide_image (Gemini): 5 seeds per image

  • add_soundtrack (ElevenLabs Music): 5 seeds per track

  • add_narration (ElevenLabs Voice): 5 seeds per batch

  • animate_slide (image-to-video, OPTIONAL): 10 seeds per live clip. Animates the subject inside a still (blink, breath, micro-gesture) while the FFmpeg camera motion layers on top. A per-slide upgrade applied after you review the cinematic baseline — never by default. Cache hits via search_live_library are free.

Seeds are sold at $5.99 for 100 seeds (standard pack) or $0.50 for 5 seeds (mini trial pack). 50 seeds are granted free on signup, and up to 150 once you connect a social account.

Generation as a workflow step

Generation is never the destination — it fills in what the library can't, inside a reel session. Always search the library first (free); generate only when there's no good match:

  • search_visual_library / get_music_library — search the existing royalty-free library first. FREE. Matches at score ≥ 40 generally beat fresh AI generation.

  • add_slide_image — fill a slide with a Gemini Imagen still, 5 seeds. Workshop the prompt free via suggest_prompt({kind:"image"}) first.

  • add_soundtrack — add an ElevenLabs Music track to the reel, 5 seeds. Workshop via suggest_prompt({kind:"music"}) (free) first.

  • add_narration — narrate slides via ElevenLabs Voice, 5 seeds per batch. Pick a voice free via list_voices first.

Generated assets persist to your asset library, so they're reusable across future reels and discoverable on later searches.

Manual install (without the plugin)

If you only want the MCP and don't want the skills:

claude mcp add --transport http poppify https://poppify.ai/mcp

You'll get the same tool surface, but Claude won't have the pre-loaded knowledge about when to call which tool or how to verify the finished video.

Updates

The plugin uses semantic versioning. To get updates:

/plugin marketplace update poppify

Skill content evolves as the MCP surface evolves — we publish new versions when a tool surface changes meaningfully.

Support

License

MIT — see LICENSE.

Related MCP Connectors

Related MCP Servers