Skip to main content
If the agent you’re using doesn’t have an official TapKit plugin, skills are how you bring TapKit knowledge in. Skills are Markdown files that teach AI agents how to use apps on your phone — they follow the open Agent Skills standard and work with any agent that supports them. You should always use skills alongside one of TapKit’s connection methods — the MCP server or the CLI. Skills tell the agent what to do; the MCP server and CLI give it the ability to do it.

​
Install skills

npx skills add jootsing-research/skills
This installs skills into your project’s .agents/skills/ directory, where most agents automatically pick them up.

​
Available skills

​
TapKit (core)

The core tapkit skill teaches the agent how to use TapKit itself — the coordinate system, screenshot loop, gesture types, and how to chain actions together. Always install this one.

​
App skills

App skills map out a specific iOS app’s UI — tabs, buttons, navigation patterns, and gotchas. They tell the agent what it’s looking at when it sees a screenshot.
SkillAppWhat it covers
tapkit:hingeHingeProfile cards, roses, messaging, preferences
tapkit:telegramTelegramChat types, reactions, search, bots, groups
tapkit:tiktokTikTokFeed, video actions, DMs, creator pages
tapkit:linkedinLinkedInFeed, networking, jobs, messaging
tapkit:uber-eatsUber EatsRestaurant search, ordering, checkout, reorders
tapkit:twitterTwitter / XComing soon
tapkit:instagramInstagramComing soon
Install the core skill plus whichever app skills match your use case.

​
Using skills with a connection method

Skills on their own don’t give the agent access to your phone — they just teach it how apps work. You need to pair them with a connection method:
  • MCP server — connect TapKit as a remote MCP server so the agent can execute taps, swipes, screenshots, and other actions
  • CLI — install the TapKit CLI and let the agent run commands directly
You can also build your own integration on top of the API.