Agent Tools & Commands

Definition: Agent tools are the actions an AI agent can take in the real world beyond chatting. When you ask an agent to find a price, file a task, email a client, or trigger an automation, it picks the right tool, calls it, and uses what comes back. In Taskade, every agent ships with 34 built-in tools for AI Agents, so it works inside your projects from the first message.

This is the difference between an agent that answers questions and one that does the work. A plain chatbot can describe how to schedule a meeting. An agent with tools books it, adds the attendees, and confirms the time. Tools turn a conversation into an outcome.

TL;DR: Agent tools (also called function calling) let an AI agent take action: web search, file handling, project and automation management, image generation, persistent memory, and more. Taskade gives every agent 34 built-in tools and connects to 100+ integrations, so non-technical teams ship working AI teammates with no code. Build one free →

You're already doing a version of this by hand. You copy a lead from your inbox into a spreadsheet, search the web for context, then update a status somewhere. An agent with tools does that loop for you, on request or on a schedule.

What Are AI Agent Tools?

AI agent tools are predefined actions an agent can call to read or change data outside the chat window. The model decides which tool fits the request, fills in the inputs, runs it, and reads the result back into its reasoning. This pattern is often called function calling. In Taskade, agents get 34 built-in tools out of the box, so they can search, write, and automate without setup.

Without tools, an agent is limited to text it can generate. With tools, it reaches into your projects, the open web, your calendar, and your automations. The agent stays in control of the conversation; the tools do the real work and report back.

How Function Calling Works

Function calling is a four-step loop: the agent reads your request, picks a tool, calls it with the right inputs, then uses the result to answer or take the next action. If one tool's output points to a next step, the agent can chain another call. You approve sensitive actions before they run.

The agent never guesses at data it can fetch. Ask "what's the latest on this client," and it searches the project, pulls recent notes, and answers from real records instead of inventing them. The loop is why agents with tools stay accurate while plain chatbots drift.

The 34 Built-In Agent Tools

Taskade gives every agent 34 built-in tools across six working categories: project and task management, web research, communication and calendar, files and documents, automation and execution, plus memory and reasoning. You enable only the tools an agent needs, so each teammate stays focused on its job.

Tools draw on the same 100+ integrations that power Taskade automations. Triggers pull events in, actions push data out, so an agent can read a new email and update a project in one move. For deeper recall across sessions, pair tools with agent memory.

A picker's-eye view of the toolbox

┌──────────────────────────────────────────────────────┐
│  AGENT TOOLBOX  ·  34 built-in tools                   │
├──────────────────┬───────────────────────────────────┤
│  Project & task  │  ▸ create / update / complete task │
│                  │  ▸ read & search projects          │
├──────────────────┼───────────────────────────────────┤
│  Web research    │  ▸ web search   ▸ web browse        │
│                  │  ▸ extract data ▸ analyze content   │
├──────────────────┼───────────────────────────────────┤
│  Comms & cal     │  ▸ email   ▸ messages   ▸ events    │
├──────────────────┼───────────────────────────────────┤
│  Files & docs    │  ▸ create / find / copy / process   │
├──────────────────┼───────────────────────────────────┤
│  Automation      │  ▸ run flows ▸ HTTP ▸ code ▸ images │
├──────────────────┼───────────────────────────────────┤
│  Memory          │  ▸ persistent recall ▸ retrieval    │
└──────────────────┴───────────────────────────────────┘

The agent reads this menu the way you'd scan a list of buttons. It matches your request to the closest action, then calls it.

Integration Examples

Connected tools let an agent act inside the apps your team already uses. Two of the most common are calendar and file management, both handled in plain language. You describe the outcome, the agent maps it to the right tool, and you approve the action before it runs.

Calendar

• Create a quick event with smart defaults (1-hour duration, 10-minute reminder)
• Schedule a detailed event with specific times, locations, and attendees
• Update, reschedule, or cancel existing events

Example commands:

• "Create a calendar event 'Team Meeting' on Friday at 3 PM"
• "Add a quick event 'Client Call' tomorrow at 2 PM"
• "Schedule weekly standups for the next month"

Files & documents

• Create, duplicate, search, and organize files and folders
• Generate new documents from text input or a template
• Share files and manage access

Example commands:

• "Create a file titled 'Monthly Reports' in my Drive"
• "Make a new folder named 'Project Alpha'"
• "Find all documents with 'budget' in the title"

Adding Tools to an Agent

You add tools by editing an agent and turning on the ones it needs. There's no code and no configuration file. Open the agent, go to the Tools tab, and enable each tool by name. Save, and the agent can use them in its next reply.

1. Open Agents in your workspace or folder sidebar
2. Select an agent (or create a new one)
3. Edit the agent from the arrow next to its name
4. Enable tools on the Tools tab, then click Add next to each
5. Save to activate

Once enabled, you can talk to the agent three ways: from the chat balloon inside any project, from the Ask AI button in the corner, or from a dedicated chat on the Agents tab. In every mode you review and approve actions before they run, so you stay in control. For step-by-step screenshots, see the Learn guide to agent tools.

Custom Commands & Control

Beyond the built-in set, you can give an agent custom slash commands that bundle a prompt and a set of tools into one reusable action. Pair these with mode selection: Default mode answers prompt by prompt, while Plan and Execute mode pursues a goal across several tool calls. See custom commands and slash commands for the full setup.

Control stays with you at every layer:

• Approval gates require confirmation before sensitive actions run
• Scoped access means each agent only holds the tools you grant it
• Activity history logs what each tool did, so you can audit any action
• Human in the loop keeps a person on approvals for high-stakes steps, covered in human in the loop

For agents that coordinate, orchestration shows how multiple tool-equipped agents hand work to each other, and autonomous agents covers the goal-driven end of the spectrum.

Tools vs Skills vs Memory

Tools, skills, and memory are three layers of agent capability. Tools are the actions an agent can take. Skills are reusable procedures built on top of those actions. Memory is what the agent remembers between turns. Together they make a teammate that acts, follows a method, and improves with context.

What You'd Build in Taskade

Picture an ops dashboard that runs your week without you copying data between tabs. You describe it in plain English to Taskade Genesis, and Taskade EVE, the meta-agent behind Taskade Genesis, builds the app and wires up an agent with the right tools. Each morning the agent searches for fresh updates, reads your active connected projects, drafts the day's tasks, and posts a summary to your team channel. You open one screen, see what changed overnight, and approve the few items that need a human. The agent did the gathering; you keep the decisions.

That's the shift tools make: the work that used to live across your inbox, a spreadsheet, and your memory now runs in one place, on its own. Build your first agent free →

Related concepts: AI Agents · Agent Memory · Agent Skills · Custom Commands · Slash Commands · Autonomous Agents · Learn: Agent Tools