Ask Claude. Get the video.

Connect the Percify MCP server once. Your avatar, your voice and 120+ image, video and voice models then answer from inside the chat you already use.

Connect in Claude

Sign in with your Percify account · no API key

Works inClaudeChatGPTCursorClaude CodeCodex
Claude
Make a 30 second LinkedIn video of me explaining why we moved our onboarding to video. My cloned voice, captions on.

I will draft the script from your notes, render it on your avatar in your voice, then burn in captions for a vertical cut you can post today.

Percify
Composed the video9:16 · 0:30 · Captions · Your voice
A finished LinkedIn video, captions burned in. Create in Claude
Result

Connected in a minute

1

Copy the server URL

https://mcp.percify.io/mcp

You will paste it in the next step.

2

Add the connector

In Claude open Settings, then Connectors, then Add custom connector. Name it Percify and paste the URL.

Open Claude connectors
3

Connect and sign in

Click Connect. A Percify sign in opens; log in and press Authorize. Then ask Claude for a video.

Claude Code

claude mcp add percify --transport http https://mcp.percify.io/mcp --header "Authorization: Bearer pk_live_KEY"

Cursor, Codex, any MCP client

{ "mcpServers": { "percify": {
  "url": "https://mcp.percify.io/mcp",
  "headers": { "Authorization": "Bearer pk_live_KEY" } } } }

Keys are only for clients without a browser sign in. Create one at app.percify.io, Developer · Setup for every client and the full tool list in the MCP server docs · REST reference at docs.percify.io

What your agent can do

55 tools, one account, one credit meter. Read only tools approve themselves; anything that spends credits asks first.

Your avatar, speaking

Save a face and a voice once. From then on "make my avatar say…" returns a lip synced video, with a preview rendered right in the chat.

create_avataravatar_sayget_avatar_joblist_avatarsdelete_avatar
Make my avatar say the intro for episode 12.

A finished short from a topic

One topic becomes a script, a presenter, a voice, captions and a retention edit. Recut the edit without rendering the lip sync again.

make_videoreedit_videoget_mascot
Make a 40 second short on why we moved onboarding to video.

Replicate a viral video

Paste a link. The shot list, beats and captions come back as a blueprint, then a replica renders clip by clip with a side by side compare.

analyze_videoget_video_analysisreplicate_videoget_replication
Analyze this Reel and remake it with my product in it.

Any model, priced first

120+ image, video and voice models behind one contract. Estimate before you spend, wait server side, check the balance, cut a model versus model ad.

list_modelsget_modelestimate_costgeneratewait_for_generationget_generationcompare_modelsget_comparecheck_usage
What would a 10 second Seedance clip cost me?

Every model your agent can call

See all 120+ models
GPT Image 2
GPT Image 2imageTry
InfiniteTalk Fast
InfiniteTalk Fasttalking videoTry
Seedance 2.0 Fast
Seedance 2.0 FastvideoTry
Pixverse V6
Pixverse V6image to videoTry
Flux 2 Dev
Flux 2 DevimageTry
Hailuo 2.3
Hailuo 2.3videoTry
Nano Banana 2
Nano Banana 2image editTry
Zonos 2
Zonos 2voiceTry

Built for agents that spend money

Every line here is server behaviour, not a promise.
Sign in, no key pastedClaude connects with OAuth to your Percify account.
Priced before it runsestimate_cost returns the exact credits, and costs nothing.
Never double chargedRetries carry an idempotency key and return the original job.
Refunds on failureA failed generation gives its credits back automatically.

Before you connect

Everything you need to know before you start.

Percify runs a standard MCP server (Streamable HTTP) at mcp.percify.io/mcp. Any client that speaks the Model Context Protocol connects with that one URL. Claude signs you in with OAuth; other clients send an API key in a header.

Not in Claude. Click Connect and sign in with your Percify account. Keys are only for clients without a browser sign in, and you create them in the developer console at app.percify.io.

Claude on web, desktop and mobile through custom connectors. Claude Code, Cursor, Codex and any client that can send a header use an API key.

The same credits as the app, charged to the connected account. estimate_cost prices a run before it spends anything, failed generations refund, and retried requests never charge twice.

Images, video clips, talking avatars in your cloned voice, dubs, fully edited shorts from a topic, and replicas of existing videos with your face or your product in them.

The same server also carries the Brand OS tools (read_brand, todays_round, schedule_post, set_autopilot and more), which read a brand from its website and post for it. Brand OS is Percify’s distribution product and is documented on percify.ai. This page covers making the video.

Images finish in seconds, video in one to a few minutes. Tools return an id at once, and wait_for_generation holds the line server side until the result is ready.

Yes. Remove the connection from your Percify account at any time, or revoke the key in the developer console. A connection only ever spends the credits of the account that approved it.

PercifyTalking videos of you, from one photo

Connect Percify to Claude.

One URL and one sign in. Then your avatar answers from inside the chat.

  • 25,000+ creators
  • From one photo
  • Cancel anytime