I Have a Perplexity Pro Subscription. I Want to Turn it into a Terminal AI Like Claude Code

How I found an open-source repo that unlocked my 1-year perplexity subscription into a full-blown CLI chatbot, MCP server, and REST API

There’s a specific kind of joy that comes from finding a tool that completely changes how you interact with something you already pay for.

For me, that moment happened when I stumbled across a GitHub repository called perplexity-scrape. What started as a curious afternoon experiment ended up reshaping my entire AI workflow, and it all started with a simple question: “Why am I only using Perplexity through a browser?”

A Bit of Background: What Is Perplexity?

Perplexity is an AI-powered search engine and answer engine that provides conversational responses to user queries by reading and synthesizing information from the live web [optus, 2025]

The Pro subscription is around $20/month, and lucky me, I got a 1-year trial.

The Problem with the Browser

Perplexity’s web interface is okay but limited, and if you live in your terminal the way I do, switching to a browser window breaks your flow. You lose context. You lose history. You can’t easily pipe answers into other tools or scripts. I wanted something like Claude Code or Codex CLI, the tools where we can open a terminal, type a question, and get a rich, contextual AI response without ever touching a GUI.

Building the CLI: Turning Perplexity Into a Terminal Companion

After forking the repo, I added a cli.py that brings together everything into a rich interactive experience. Here's what it ended up looking like:

screenshot of perplexity-cli

Starting the CLI

# Basic — starts with the default model (Claude 4.5 Sonnet Thinking)
python3 cli.py

# Pick a specific model
python3 cli.py --model gpt-5.2

# Load MCP tools (e.g., BurpSuite, filesystem access)
python3 cli.py --mcp ./mcp_tools.json

# Resume a previous conversation
python3 cli.py --load ~/.perplexity_cli/sessions/session_20250524.json

# Go incognito (doesn't appear in your Perplexity dashboard)
python3 cli.py --incognito

# Set a custom system prompt
python3 cli.py --system "You are a security researcher. Focus on vulnerability analysis."

Slash Commands Inside the CLI

Once you’re inside the CLI, a full set of slash commands gives you control over every aspect of your session:

/model <name>   #Switch AI model mid-conversation
/models         #Browse all available models
/mode <name>    #Switch between copilot, search, reasoning, writing
/sessions       #List all saved sessions
/tools          #Show connected MCP tools
/status         #Session info at a glance

The MCP Integrations

By providing a mcp_tools.json config file, you can connect the CLI to any MCP-compatible tool server, and the AI model will automatically use them when relevant.

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/home/user"]
    },
    "github": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-github"
      ],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "<YOUR_TOKEN>"
      }
    }
  }
}

Then start the CLI with:

python3 cli.py --mcp ./mcp_tools.json

Now, when you ask “list the files in my project directory,” the model doesn’t just answer generically; it actually calls the filesystem tool and returns real results from your machine.

Screenshot of MCP tools integration with Perplexity-cli

Setting It Up: 10 Minutes to Your Own (perplexity) AI Terminal

1. Clone the repo :

git clone https://github.com/mrdzer0/perplexity-scrape.git
cd perplexity-scrape
git switch mcp-cli-enh
pip install -r requirements.txt

2. Get your Perplexity session token: Log into Perplexity in your browser, then open DevTools → Application → Cookies → find perplexity.ai → copy the __Secure-next-auth.session-token value.

3. Create your .env file:

PERPLEXITY_SESSION_TOKEN=your_token_here

4. Launch:

# Interactive CLI
python3 cli.py

# Or the full unified server (REST API + MCP HTTP)
python unified_service.py

That’s it. Now I’m running Frontier AI models from the terminal using a subscription that I get from Perplexity.

Last but not least, enjoy the vibe coding :)

Tags: #AI #Perplexity #CLI #MCP #OpenSource #Terminal #Python #DeveloperTools #LLM