Skip to content
tellahqPublic

About

No description, website, or topics provided.

Resources

Stars

2 stars

Watchers

0 watching

Forks

Latest commit

 

History

38 Commits

Folders and files

Repository files navigation

Tella Agent Skills

skills.sh

Agent Skills for working with Tella videos from Claude Code, Codex, Cursor, Antigravity, and any other skills-compatible agent.

npx skills add tellahq/skills

Claude plugin

This repository is also the Tella plugin for Claude. The plugin bundles every skill below with the Tella MCP server, so Claude can find, analyze and edit your Tella videos without any setup beyond signing in.

In Claude Code:

/plugin install tella --marketplace tellahq/skills

In claude.ai and Cowork, install it from the plugin directory, then connect Tella on the plugin's Connectors tab. The plugin uses the same server as the Tella connector, so if you've already connected it you'll see one set of Tella tools, not two. In Claude Code, sign in from /mcp or the first time Claude uses a Tella tool. Tella opens in your browser so you can sign in and pick a workspace.

Cursor plugin

The same repository is the Tella plugin for Cursor, with the same skills and the same Tella MCP server. Install it from the Cursor Marketplace, or try it before it's listed by copying this repository into ~/.cursor/plugins/local/tella and reloading Cursor. Cursor asks you to sign in to Tella the first time it uses a Tella tool.

Antigravity plugin

The same repository is also the Tella plugin for Google Antigravity, with the same skills and the same Tella MCP server. Install it from GitHub in the Antigravity CLI:

agy plugin install https://github.com/tellahq/skills

Plugins installed with agy also show in the Antigravity App under Customizations > Installed. Antigravity asks you to sign in to Tella in your browser the first time the Tella server connects.

What the plugins connect to

Every plugin adds one remote MCP server, tella, at https://api.tella.com/mcp (Streamable HTTP, signed in with your Tella account over OAuth). There are no hooks or local servers. One skill, tella-audio-sync, includes a Python script, kept as text in the skill's Markdown, that the agent writes out and runs on your machine when you ask it to fix lip sync: it downloads short clips of your camera recording from Tella and measures them locally, after installing mediapipe 1.2.2, scipy 1.16.3 and numpy 2.3.5 (plus their dependencies) from PyPI and downloading version 1 of MediaPipe's face landmark model from Google (storage.googleapis.com/mediapipe-models/face_landmarker/face_landmarker/float16/1/face_landmarker.task, checked against its SHA-256 before use). Every other skill is instructions only.

When Claude, Cursor or Antigravity calls a Tella tool, the tool's inputs (such as a video or clip id, a time range, overlay text or a script) go to Tella, and Tella returns data from the videos your account can open: titles, transcripts, timelines, frames, previews and analytics. Edits change your videos in Tella. Nothing is sent anywhere else. Tella handles that data under its privacy policy.

Skills

tella

Analyze, edit, and visually verify Tella videos through the Tella MCP server: trims, layouts, overlays, zooms, captions, format changes, and export. The skill explains how to connect the MCP if it is not set up yet.

Example prompt: /tella Tighten the pacing of my latest video and make it 9:16 for mobile

tella-remove-mistakes

Find recording mistakes in a Tella video (false starts, retakes, repeated phrases, abandoned sentences, technical interruptions), review them with you, and cut the approved ones through the Tella MCP.

Example prompt: /tella-remove-mistakes Clean up the retakes in my latest video

tella-auto-layouts

Lay out a Tella clip automatically: pick a base layout, then add layout changes (screen focus, camera cutaways, punch-ins, bubble size changes) only where they help the viewer, keeping the clip's camera style and pop-out.

Example prompt: /tella-auto-layouts Lay out my latest video as a product demo

tella-auto-cut

Tighten a Tella video in one go: cut recording mistakes, filler words and silent pauses, then clean up the gaps and broken seams the cuts leave behind.

Example prompt: /tella-auto-cut Tighten up my latest video

tella-short

Turn a video into a vertical Short, Reel or TikTok: pick its strongest 15 to 60 seconds, make a trimmed 9:16 copy, tighten it, add captions, and lay it out at Shorts pace with a hook up front, B-roll that moves and the speaker on screen. It shows a scene-by-scene plan first.

Example prompt: /tella-short Make a Short from the best part of my latest video

tella-audio-sync

Fix a voice that's out of sync with the speaker's lips: a Python script in the skill measures how far the voice is ahead or behind on short downloads of the camera recording, timing every moment the lips part against the voice, and the skill moves the clip's voice by that much. It needs an agent that can run Python (Claude Code, Codex, Cursor, the Tella editor). Audio sync is in beta and not available in every workspace.

Example prompt: /tella-audio-sync The audio is a bit behind my lips in my latest video, can you fix it?

tella-plan-video

Plan a video before you record it: what it's for, the one point, a word-for-word script sized to the length, and a storyboard of what's on screen at each line. It can also turn a long, loose take into a tight script to record again.

Example prompt: /tella-plan-video Help me plan a 45-second launch video for our new feature

tella-launch-video

Make a launch, promo or feature video: it asks what's launching and what material you have, picks a shape (founder to camera, product walkthrough, music-only trailer, vertical feature drop), plans it, then cuts, lays out, animates the product, adds music and an end card.

Example prompt: /tella-launch-video Turn my recording into a launch video for our landing page

tella-motion-design

The shared rules the motion graphics skills below follow: one look per video (typeface, palette, easing, timing), graphics timed to the spoken word, reading limits, clichés to avoid, which Tella tool builds which graphic, and checking the result frame by frame.

Example prompt: /tella-motion-design Set up a motion style for my latest video that matches my brand

tella-kinetic-type

Put the lines that matter on screen as animated type that lands on the spoken word: emphasis words beside the speaker, list builds, big titles behind the presenter, or a full-frame hook.

Example prompt: /tella-kinetic-type Add kinetic type to the key moments of my latest video

tella-title-cards

A matching set of titles from one design: an opening title, chapter titles, name tags and an end card.

Example prompt: /tella-title-cards Add an opening title and chapter cards to my latest video

tella-explainer-graphics

Turn an explanation into a diagram that builds itself as the speaker talks, with parts appearing as they are named and arrows moving the way things flow.

Example prompt: /tella-explainer-graphics Visualize the architecture I explain at 2:10

tella-data-callouts

Show the numbers the speaker says as big stats, count-ups, bars and comparisons, each timed to the spoken number and checked against the transcript.

Example prompt: /tella-data-callouts Visualize the results I mention in my latest video

tella-screen-callouts

Point at things in a screen recording: spotlights, labels with arrows, and numbered steps, placed on the UI and timed to the click or the words that mention it.

Example prompt: /tella-screen-callouts Annotate the setup steps in my demo

tella-ui-redraw

Redraw your product's real screens natively in Tella Motion, as live shapes and text simplified to what the shot needs, then animate the UI doing its thing: the cursor clicks, a panel opens, a list fills, a number counts up.

Example prompt: /tella-ui-redraw Make animated UI shots of the dashboard steps in my demo

tella-ui-parallax

Turn a real screenshot into a parallax shot: cut its pieces out, lift the one that matters towards the viewer, blur the rest back, and move one camera through the layers. Nothing is redrawn, so it is exactly your product.

Example prompt: /tella-ui-parallax Make a parallax shot of the invoice card in this screenshot

tella-ui-interactions

Recipes for UI that acts in Tella Motion: presses, menus and modals growing from a click, toggles, tabs, typing, loading, success checks, notifications and counters, each with its timing.

Example prompt: /tella-ui-interactions Animate the settings toggle and the save confirmation

tella-carousels

Show several images, screens or clips as a carousel (a ring, an arc, a row or a stack) that lands on the one that matters as it's named.

Example prompt: /tella-carousels Show our five templates as a carousel when I mention them

tella-transitions

Pick transitions that fit each cut: Tella's clip, layout and zoom transitions, or ones built in Tella Motion from something on screen (zoom through, grow into, wipe, match cut, whip), checked for flash frames.

Example prompt: /tella-transitions Make the topic changes in my latest video feel smoother

tella-camera-moves

Zooms, punch-ins and camera moves with a reason and a hold: zoom to read, follow the cursor, punch in on a key line, push in, pull back to reveal, pan, focus pulls.

Example prompt: /tella-camera-moves Add zooms where I click through the settings

tella-changelog-video

Turn release notes into a short update video, one beat per change shown in the product, ready to repeat every release.

Example prompt: /tella-changelog-video Make an update video from these release notes

The motion graphics skills use text overlays where they can. Diagrams, charts, count-ups, arrows and the richer type styles use the Tella Motion tools, which are in beta and not available in every workspace.

About

No description, website, or topics provided.

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors