Dexscreener Unofficial CLI + MCP + Skills

100% free to use. All APIs included are public and free - no Dexscreener API key required to get started. Optional free Moralis key unlocks holder data.

A visual terminal scanner, MCP server, and AI skill for Dexscreener token signals. Unofficial - not affiliated with or endorsed by Dexscreener.

Scans hot tokens across every chain Dexscreener supports. Scores them by volume, liquidity, momentum, and flow pressure. Use it from the terminal, connect it to AI agents via MCP, or load it as a skill in Claude/Codex/OpenClaw.

---

TL;DR - Get scanning in 60 seconds

Windows:

git clone https://github.com/vibeforge1111/dexscreener-cli-mcp-tool.git
cd dexscreener-cli-mcp-tool
install.bat
ds setup
ds hot

Mac / Linux:

git clone https://github.com/vibeforge1111/dexscreener-cli-mcp-tool.git
cd dexscreener-cli-mcp-tool
chmod +x install.sh && ./install.sh
./ds setup
./ds hot

That’s it. After install you can just type ds (Windows) or ./ds (Mac/Linux) from the project folder. No activation, no paths to remember.

---

What’s Installed

One install gives you everything:

Think of it this way:

• CLI = best live experience and fastest workflows
• MCP = lets an AI agent operate the scanner for you
• Skill = teaches the agent when to use MCP vs sending you to the CLI

---

Install

You need Python 3.11+ and Git. That’s it.

Step 1: Clone the repo

Open a terminal and paste this:

git clone https://github.com/vibeforge1111/dexscreener-cli-mcp-tool.git
cd dexscreener-cli-mcp-tool

Step 2: Run the installer

Windows:

install.bat

Mac / Linux:

chmod +x install.sh && ./install.sh

This creates a virtual environment, installs all dependencies, and runs diagnostics automatically. Takes about 30 seconds.

Step 3: Set up your preferences

ds setup

Asks you 5 quick questions (which chains, trading style, etc.) and saves your defaults. Takes 30 seconds.

Step 4: Start scanning

ds hot

That’s it. You’re scanning.

Running commands

After install, short wrapper scripts let you skip the .venv path entirely:

You can always use the full path if you prefer: .\.venv\Scripts\ds.exe (Windows) or ./.venv/bin/ds (Mac/Linux).

---

Your First Live Dashboard

If you only run one more command after setup, make it this:

Windows:

ds new-runners-watch --chain=solana --watch-chains=solana,base --profile=discovery --max-age-hours=48 --include-unknown-age --interval=2

Mac / Linux:

./ds new-runners-watch --chain=solana --watch-chains=solana,base --profile=discovery --max-age-hours=48 --include-unknown-age --interval=2

This is the best live mode. It auto-refreshes, shows rank movers and spotlight cards, and lets you switch chains with 1/2 keys. Press Ctrl+C to stop.

Common Windows mistake

If you see Option '--profile' requires an argument, you pressed Enter too early. The whole command needs to be on one line. Using = signs (like --profile=discovery) helps because the option and value stay glued together.

---

What Can I Do With This?

“I just want to see what’s hot right now”

ds hot

“I only care about Solana”

ds hot --chains solana --limit 10

“Show me tokens on Base too”

ds hot --chains solana,base --limit 15

“I want a live feed that updates automatically”

ds new-runners-watch --chain=solana --watch-chains=solana,base --profile=discovery --max-age-hours=48 --include-unknown-age --interval=2

“Show me brand new tokens that just launched”

ds new-runners --chain solana
ds top-new --chain base

“Find me alpha - new drops with breakout potential”

ds alpha-drops --chains solana,base

“Search for a specific token”

ds search pepe
ds search 0x6982508145454ce325ddbe47a25d4ec3d2311933

“I found a token, give me everything on it”

ds inspect So11111111111111111111111111111111111111112 --chain solana

“Alert me on Discord when something hot appears”

ds task create my-alerts --preset scout --interval-seconds 60
ds task configure my-alerts --discord-webhook-url https://discord.com/api/webhooks/YOUR/WEBHOOK
ds task test-alert my-alerts
ds task daemon --all

“I want JSON output for my own scripts”

ds hot --json
ds hot --chains solana --limit 5 --json > tokens.json

---

Commands

Scans

Live Dashboards

Three live modes that auto-refresh and keep your terminal updated. Press Ctrl+C to stop any of them.

Tips for all live modes:

• --interval 5 for slower, less chatty refreshes (default is 2 seconds)
• --limit to control how many tokens show (fewer = faster)
• --profile discovery to cast a wider net
• 1/2 keys to switch chains (if --watch-chains is set)
• s to cycle sort mode, j/k to navigate, c to copy address

Setup & Maintenance

---

MCP Server - Use It With AI Agents

The MCP server lets you talk to your scanner in plain English through any AI agent. Instead of remembering CLI flags, you just ask:

• “What’s pumping on Solana right now?”
• “Find me degen plays on Base with low liquidity”
• “Save a preset called sol-degen for Solana discovery mode”

How to set it up

Step 1: Make sure the CLI is installed (see Install above).

Step 2: Add the MCP server to your AI agent’s config.

Claude Desktop - add to your claude_desktop_config.json:

Mac/Linux:

{
  "mcpServers": {
    "dexscreener": {
      "command": "/path/to/dexscreener-cli-mcp-tool/.venv/bin/dexscreener-mcp",
      "args": []
    }
  }
}

Windows:

{
  "mcpServers": {
    "dexscreener": {
      "command": "C:\\path\\to\\dexscreener-cli-mcp-tool\\.venv\\Scripts\\dexscreener-mcp.exe"
    }
  }
}

Claude Code - add to your .mcp.json or project settings:

Mac/Linux:

{
  "mcpServers": {
    "dexscreener": {
      "command": "/path/to/dexscreener-cli-mcp-tool/.venv/bin/dexscreener-mcp"
    }
  }
}

Windows:

{
  "mcpServers": {
    "dexscreener": {
      "command": "C:\\path\\to\\dexscreener-cli-mcp-tool\\.venv\\Scripts\\dexscreener-mcp.exe"
    }
  }
}

Any MCP-compatible agent (Codex, OpenClaw, etc.) - point it at the dexscreener-mcp binary in the .venv folder. It communicates over stdio.

Step 3: Start talking.

---

AI Skill File

For AI coding agents that use skill files (Claude Code, Codex, OpenClaw), load SKILL.md from this repo. It teaches the agent when to activate, how to map natural language to tool calls, chain identification, filter selection, score explanation, and error handling.

---

---

Environment Variables

For disclosure and reporting guidance, see SECURITY.md.

---

Troubleshooting

Run ds doctor anytime to check your setup.

---

Updating

ds update

Or manually:

git pull
pip install -e .

---

Project Structure

dexscreener_cli/
  cli.py          - All CLI commands (Typer)
  ui.py           - Terminal rendering (Rich)
  scanner.py      - Token discovery and scanning
  scoring.py      - 8-component scoring engine
  models.py       - Data models (PairSnapshot, HotTokenCandidate)
  holders.py      - Multi-provider holder count fetching
  client.py       - Dexscreener API client with rate limiting
  config.py       - Constants and filter defaults
  state.py        - Preset/task persistence (JSON files)
  mcp_server.py   - MCP server exposing all tools
  alerts.py       - Discord/Telegram/webhook alerts
  task_runner.py   - Task execution and scheduling
  watch_controls.py - Keyboard controls for live mode

---

License

MIT