How to Use Pi Agent: Install, Log In, Choose a Model & Automate Workflows with Skills in 5 Minutes
Want to know how to use Pi Agent? Pi Agent is a general-purpose AI agent built on a minimal core — you extend its capabilities by installing Skills on demand. It focuses on automating everyday workflows (searching for information, reading PDFs, organizing spreadsheets, making PPTs) rather than writing code. Whatever Skills you install, that's what it can do. This article takes you from installation to a real-world demo in 5 minutes.
Table of Contents
- Prerequisites: What You Need Before You Start
- What Is Pi Agent and Why Should You Use It?
- What's the Difference Between Pi Agent and Claude Code?
- Experience Pi Agent in 5 Minutes
- Hands-On: Make a PPT with Pi Agent
- Pi Design Philosophy: Minimal Core + On-Demand Expansion
- Quick Reference: Common Keyboard Shortcuts
- Common Slash Commands
- FAQ: Frequently Asked Questions About Using Pi Agent
Prerequisites: What You Need Before You Start
Before learning how to use Pi Agent, make sure your environment meets these requirements:
- Node.js 20+ (latest LTS recommended)
- A working AI model API Key (e.g., DeepSeek, Claude, OpenAI) or a subscription account (Claude Pro, ChatGPT Plus, etc.)
- A project directory (recommended: launch Pi from the project root so it can read your local files)
Once these are in place, you're ready to walk through the full how-to-use-Pi-Agent workflow.
What Is Pi Agent and Why Should You Use It?
On the OpenRouter leaderboard, Pi ranks fifth in daily token consumption, right behind OpenClaw.
Even more telling: Pi runs on "low consumption, high frequency." For tasks of the same scale, Pi consumes roughly a third (or less) of the tokens Claude Code does, yet its total consumption still lands in the top six. That means a lot of users rely on it for daily work.
The one-sentence value of using Pi Agent: it's not a tool for writing code — it's a general-purpose agent that automates everyday chores like searching for information, reading PDFs, tidying spreadsheets, writing reports, and building PPTs.
What's the Difference Between Pi Agent and Claude Code?
People searching "how to use Pi Agent" usually first ask how it differs from Claude Code. This table makes it clear:
| Dimension | Claude Code | Pi Agent |
|---|---|---|
| Core purpose | Writing code and shipping software | Getting everyday tasks done |
| Preinstalled capabilities | A full dev toolset: code indexing, test running, Git operations, coding standards, etc. | Only four basic abilities: read files, write files, modify files, run commands |
| Extension method | MCP / Hooks | Skills installed on demand |
| Target users | Developers | Anyone with everyday office needs |
| Typical scenarios | Building features, fixing bugs, code review | Searching info, reading PDFs, organizing spreadsheets, writing reports, making PPTs |
The core difference: Claude Code solves development efficiency; Pi solves workflow efficiency. Not everyone writes code every day, but everyone has a pile of chores to handle. That's why more and more people are searching "how to use Pi Agent".
Experience Pi Agent in 5 Minutes
Here's the step-by-step guide to actually using Pi Agent.
Step 1: Install Pi Agent
Run this command in your terminal:
curl -fsSL https://pi.dev/install.sh | sh
After installation, all configs, session records, and Skill packages live in the ~/.pi/agent/ directory.
Step 2: Launch and Log In
Start Pi from your project directory:
pi
Type /login to log in. You have two options:
Option 1: Subscription login (if you already have Claude Pro, ChatGPT Plus, etc.)
Type /login, choose a provider, and follow the prompts to authorize.
Option 2: API Key (recommended for easy secret management)
Step 3: Choose and Configure a Model
After logging in, type /model or press Ctrl+P to open the model picker. Press Shift+Tab to toggle thinking level.
Using DeepSeek as an example: select DeepSeek, then paste your DeepSeek API key and start chatting.
Step 4: Start Chatting
Configuration done — you can now talk to Pi directly. Congratulations! You've mastered the basics of using Pi Agent.
Hands-On: Make a PPT with Pi Agent
Install a PPT Skill, describe what you need in one sentence, and Pi generates a full PPT from scratch. It's the classic scenario that best shows how to automate work with Pi Agent.
Install the PPT Skill
pi install git:github.com/op7418/guizang-ppt-skill
Generate a PPT with One Sentence
Just tell Pi what you want:
Make a PPT introducing the scenery of Nanjing
Pi will analyze the request and lay out an outline. You confirm the style and accent colors, then it starts generating.
The Final Result
Below is the plain-text version generated directly (for better visuals, you'd refine the prompt and parameters further).
Pi Design Philosophy: Minimal Core + On-Demand Expansion
Pi Agent's core approach is the exact opposite of other agents: everyone else adds; Pi subtracts.
Its base only ships four fundamental abilities: read files, write files, modify files, run commands. Nothing else is preinstalled.
Why design it this way? Because Pi wants you to build your own agent:
- Researchers → install PDF reading and search Skills
- Office workers → install spreadsheet Skills
- Video creators → install Hyper Frames
- Want it to speak → install Edge TTS
- Want image assets → install GPT Image 2
Install one Skill and your agent gains one ability. Every person's Pi ends up looking different.
The quote on Pi's official site nails it:
There are many different types of agents in the world, but this Pi Agent is your own agent.
Core philosophy:
Adapt Pi to your workflows, not the other way around
That's the deepest understanding of "how to use Pi Agent": it's not a fixed-function tool but a work assistant that keeps growing to fit your needs.
Quick Reference: Common Keyboard Shortcuts
Master these shortcuts to use Pi Agent more efficiently:
| Feature | How |
|---|---|
| File reference | Type @ to fuzzy-search project files |
| Path completion | Press Tab to complete paths |
| Multi-line input | Shift + Enter (Ctrl + Enter in Windows Terminal) |
| Paste images | Ctrl + V (Windows: Alt + V), or drag an image into the terminal |
| Run shell commands | !command (output is sent to the model) |
| Hide shell commands | !!command (output is not sent to the model) |
| External editor | Ctrl + G opens $VISUAL or $EDITOR |
Common Slash Commands
Type / in a conversation to open the command-completion menu. Here are the most useful commands for how to use Pi Agent:
| Command | Description |
|---|---|
/model | Switch models |
/settings | Configure thinking level, theme, message transport, and protocol |
/resume | Pick and continue a previous session |
/new | Start a new session |
/name <name> | Set the session display name |
/session | Show session file, ID, message count, token count, and cost |
/tree | Jump to any node in the session tree and continue |
/fork | Create a new session from a previous user message |
/clone | Copy the current active branch to a new session |
/compact [prompt] | Manually compact the context (optional custom instructions) |
/copy | Copy the last assistant message to the clipboard |
/export [file] | Export the session as HTML |
/reload | Reload hotkeys, extensions, skills, prompts, and context files |
/hotkeys | Show all keyboard shortcuts |
/changelog | Show version history |
/quit | Exit Pi |
FAQ: Frequently Asked Questions About Using Pi Agent
How do I use Pi Agent, and how is it different from Claude Code?
Claude Code is for developers writing code; it ships with code indexing, testing, Git, and other dev tools. Pi Agent is for everyone handling daily tasks; its core is minimal and capabilities come from installing Skills. See the comparison table above.
Is Pi Agent free? How do I use it without wasting money?
Pi Agent itself is an open-source tool and doesn't cost anything. But you need to provide your own model API Key or subscription account; model-call fees are charged by the respective model providers.
Which models does Pi Agent support, and how do I switch?
Pi supports many models. Use the /model command to view and switch. Common options include DeepSeek, Claude models, and OpenAI models, all configurable.
Why doesn't my Skill work after installation, and how do I fix it?
Confirm the install path is correct, then run /reload to reload. If it still doesn't work, check whether the Skill file exists under ~/.pi/agent/.
You've now fully mastered how to use Pi Agent. From install, login, and model selection to automating a PPT with Skills, the whole flow fits inside 5 minutes. Go try it — let Pi truly become your own workflow assistant!