Built-in skills

Every agent starts with seven system-scope Skills already installed — no setup, no Marketplace trip. They cover the work agents need most often: reading files, finding more Skills, building new ones, scheduling work, building and previewing web apps, and checking their own machine. You rarely call these by name; the agent reaches for them when the task fits.

FAR — file augmented retrieval

Makes binary files readable to the agent. FAR generates a smart summary alongside each file — PDFs, images, spreadsheets, slide decks, videos, and more — so the agent can reason over content it couldn't otherwise open.

• When it runs — you ask an agent to read or analyze a document and it hits a format it can't parse directly.
• Result — a persistent sidecar summary the agent can reference, stored next to the original in Drive.

Find Skills

Helps an agent discover and suggest new Skills when you ask it to do something it can't do yet. Instead of failing, the agent searches the open Skills ecosystem and proposes ones to install.

• When it runs — "how do I do X", "find a skill for X", or any request that needs a capability the agent doesn't have.
• Result — suggestions you can install from the Marketplace.

Skill Creator

Build, test, and improve custom Skills directly with the agent. It interviews you about what the Skill should do and when it should trigger, drafts the SKILL.md, runs test cases, and iterates until the Skill works reliably.

• When it runs — "turn this into a skill", "I do this all the time, automate it", or a request to refine an existing Skill.
• Result — a Local Skill you can keep in your workspace or publish. See Create a custom skill.

Buda Automations

Create and manage scheduled work — set an agent to run a task on a recurring or one-time basis. The agent reads phrases like "every morning" or "once a day" and sets up the schedule for you.

• When it runs — "create a daily automation", "set up a cron job", "show my scheduled tasks", or any recurring-time phrasing.
• Schedule types — repeating intervals, cron expressions (Plus and above), or a one-time run at a future time.
• Result — a managed schedule you can monitor, pause, and resume. See Automations.

App Builder

Builds a working web application from an idea, mockup, or feature request — picks a stack (React, Vite, Next.js, Vue, Svelte, and more), scaffolds or modifies the project, implements a real vertical slice, and commits after each working step.

• When it runs — "build a web app", "make a frontend", "turn this idea into an app", or a request for a full-stack prototype, dashboard, CRUD app, or authentication flow — even without naming a framework.
• Result — a working project in the agent's workspace, typically handed off to the Web Preview Skill to launch it live.

Web Preview

Launches a live browser preview for any web page, game, or frontend project the agent creates — starts the dev server, detects and verifies the port, and opens the link in Local Browser.

• When it runs — "show me a preview", "start a dev server", "give me the link", or any request to create, view, run, or preview an HTML file, game, landing page, demo, or frontend project.
• Result — a live, clickable preview link opened in Local Browser.

Machine Info

Reports the hardware specs and resource usage of the agent's cloud computer — CPU, memory, disk, OS, and installed tool versions.

• When it runs — "what are this machine's specs", "how much memory do I have", or troubleshooting a resource limit.
• Result — a plain-language readout of the sandbox environment.

Related