ATLAS - Tools & Capabilities

🧠

Memory & Context

Persistent multi-layer memory systems for storing, searching, and managing contextual knowledge across sessions.

memoryStore

Save facts, preferences, decisions, and events into persistent memory

memorySearch

Semantically search stored memories by meaning, not exact text

memoryList

Browse all saved memory items filtered by category or tag

memoryDelete

Remove outdated or irrelevant entries from the memory store

memorySnapshot

Take a complete point-in-time export of the entire memory state

🧑‍💼

Identity & User Profile

Continuously builds and updates a structured user profile — understanding who you are, how you work, and what you prefer.

identityView

Read the current user profile: name, role, company, goals, projects

identityUpdate

Update identity fields like name, location, role, and preferences

traitUpdate

Log behavioural patterns and preferences inferred from usage

traitList

View all learned behavioural traits and their confidence scores

💻

System & OS Automation

Full control of the operating system — macOS, Windows, and cross-platform. From shell commands to brightness control.

Cross-Platform

shellExec

Run any terminal command and capture output

setVolume

Get or set system volume level (0–100), mute/unmute

clipboard

Read from or write to the system clipboard

systemInfo

Get battery, RAM, CPU, disk, Wi-Fi, and uptime snapshot

wifiControl

Turn Wi-Fi on/off, check status, connect to networks

bluetoothControl

Toggle Bluetooth, list paired devices, connect AirPods etc.

setDoNotDisturb

Enable or disable Focus/DND mode

darkMode

Toggle or set light/dark appearance mode

lockScreen

Lock the display immediately

sleepDisplay

Put the display to sleep or wake it

macOS Specific

appleScript

Execute native AppleScript to control any macOS app

osScreenshot

Capture the current screen with coordinate grid overlay

osClick / osType

Click coordinates or type text at OS level

osOcrClick

Find UI elements by visible text using OCR (Apple Vision Framework) and click them with precise jitter targeting

osVisualVerify

Confirm UI state by checking text visibility on screen

setBrightness

Set display brightness from 0 to 100

sendScreenshot

Capture and send the current screen directly to the user

osContext

Get the active app name and window title quickly

Windows Specific

powerShellScript

Execute PowerShell scripts natively on Windows

winOsScreenshot

Capture the Windows screen with a grid overlay

winOsClick / winOsType

Click or type at OS level on Windows

winOsOcrClick

Find UI elements by visible text using Windows Media OCR API and click them with precise jitter targeting

winOsVisualVerify

Verify UI state by checking on-screen text on Windows

winSetBrightness

Control Windows display brightness

winSendScreenshot

Capture and deliver the Windows screen to the user

🌐

Browser Automation

Full Playwright-powered browser control — navigation, interaction, scraping, form-filling, and reusable workflow strategies.

Navigation & Control

browserNavigate

Navigate to any URL in current or new tab

browserSwitchTab

Switch between open tabs, open new tabs, close tabs

browserClose

Close the active browser tab cleanly

browserBack

Navigate to the previous page in history

browserWait

Wait for timeout, page load, or a specific element to appear

Interactions

browserClickAt

Click elements by label ID or x/y coordinates

browserRightClick

Right-click to open context menus

browserHoldClick

Click and hold for a specified duration

browserDrag

Drag elements from one position to another

browserHover

Hover to reveal tooltips and dropdown menus

browserType

Type text into input fields by label or coordinate

browserKeyboard

Press shortcuts like Ctrl+Z, Enter, Escape, Tab

browserSelect

Choose an option from a dropdown or select element

browserMultiAction

Chain multiple browser actions in a single call

browserScroll

Scroll up/down/left/right or zoom with Ctrl+scroll

browserScrollElement

Scroll a specific element into the viewport

browserUploadFile

Upload local files to web forms and Gmail attachments

Analysis & Extraction

browserScan

Get all interactive elements as a labelled text list (or visual screenshot)

browserExtract

Pull element data using CSS selectors

browserAnalyze

Get bounding boxes and positions of all page elements

browserRunJs

Execute arbitrary JavaScript in the browser context

Strategy Memory

browserSaveStrategy

Save a successful navigation pattern for future reuse

browserRecallStrategy

Recall a previously saved browsing strategy before starting a task

📂

File Management

Full local filesystem access — read, write, search, compress, tag, and manage files and directories.

readFile

Read the full content of any file on the system

writeFile

Write or append content to a file, creates dirs if needed

listDirectory

List files and folders at a given path, recursively

searchFiles

Search for files matching a pattern across a directory

deleteFile

Permanently delete a file (with confirmation safeguard)

fileMetadata

Get/set Finder tags, Spotlight comments, creation date

compressFiles

Zip files/folders or extract existing zip/tar archives

trashManager

Move files to Trash, list trash contents, or empty it

openInApp

Open a file, folder, or URL in a specified application

sendFileTool

Send any local file to the user via Telegram

🛠

Plugins & Integrations

Deep integrations with Google Workspace, GitHub, native OS apps, media tools, 3D, and video editing.

Google Services

googleAccounts

Add, list, or remove connected Google accounts via OAuth

googleCalendarTool

List today's events, events by date, or create new calendar events

gmailTool

Read, search, send, and reply to emails across Gmail accounts

GitHub

githubTool

List repos, issues, PRs, notifications, and create new issues

Native OS Apps

spotifyControl

Play/pause/skip, search and play songs, adjust volume

musicControl

Control Apple Music — play, search, skip tracks

appManager

List running apps, quit, force-quit, hide, show, or activate

windowManage

Move, resize, split-screen, minimize, or maximize windows

calendarTool

Manage macOS Calendar — list and create events

remindersTool

Create, list, and complete macOS Reminders

messagesTool

Send iMessages and read recent conversations

runShortcut

Execute Apple Shortcuts by name or list all available

osDrag

Click-drag at OS level for native UI drag interactions

Media & Creative

generateImage

Generate images from text prompts using AI

screenRecord

Start/stop screen recording and send the video to the user

blenderScript

Execute Python against Blender's bpy API for 3D generation

davinciScript

Script DaVinci Resolve Studio for timeline and export automation

emailSend

Send emails on behalf of the user from a dedicated address

webSearch

Lightweight multi-engine web search (Google, Brave, DuckDuckGo)

🚀

Agent Operations & Task Management

Advanced orchestration — scheduling, long-running tasks, background shells, webhooks, and AI swarms.

Scheduling & Reminders

scheduleTask

Schedule recurring tasks using cron expressions

setReminder

Set a one-off reminder after a specified number of minutes

listTasks

View all currently scheduled and active tasks

cancelTask

Cancel a scheduled or queued task by ID

Webhooks & Queue

registerWebhook

Register a webhook endpoint for external service triggers

listWebhooks

List all registered webhook endpoints

taskQueueTool

Manage a sequential task queue — add, status, cancel, clear

Long-Running Operations

startLongTaskTool

Start a multi-phase background task with checkpointing

checkLongTaskTool

Monitor the status and progress of a running long task

resumeLongTaskTool

Resume an interrupted long task from its last checkpoint

cancelLongTaskTool

Cancel a running long task entirely

Background Shell

shellExecBackground

Run shell commands as background processes with a process ID

shellCheckProcess

Poll the status and output of a background process

shellWaitProcess

Block until a background process completes (up to 24h)

shellKillProcess

Forcefully kill a running background process

Advanced Agency

swarmTask

Decompose complex tasks into parallel sub-agents running simultaneously

deepResearch

Launch background multi-round research using multiple search engines and full page reads

checkDeepResearch

Check the status and results of a running deep research job

cancelDeepResearch

Stop a running deep research process

90+ Built-in Tools