Setting Up (10 Minutes, We Promise)

Five things to install. One command to test. Then you're ready to build anything on Bitcoin.

#Step 1: Install Claude Code

Claude Code is the AI agent that does the building. It's a CLI tool that runs in your terminal.

Go to claude.ai/code and follow the install instructions for your OS.

Once installed, run:

claude --version

You should see a version number. If you do, you're good.

You'll need an Anthropic account. The free tier works for exploring. For serious building, the Pro plan is worth it — you'll be running multi-agent pipelines that do a lot of work.

#Step 2: Clone the buidl-opnet-plugin

This is what transforms Claude from a generic coding assistant into an OPNet specialist. The plugin loads 8 specialized agents, OPNet-specific knowledge, and the /buidl command that runs the full build pipeline automatically.

git clone https://github.com/bc1plainview/buidl-opnet-plugin.git

Clone it somewhere permanent on your machine. You'll reference this path in the next step.

#Step 3: Set Up Your Aliases

These two aliases are how you launch Claude with the plugin loaded. Add them to your shell config (~/.bashrc, ~/.zshrc, or equivalent):

# Safe mode - Claude asks before running commands
alias claudey="claude --plugin-dir /path/to/buidl-opnet-plugin/buidl"

# Autonomous mode - Claude runs without prompting (recommended once you trust it)
alias claudeyproj="claude --dangerously-skip-permissions --plugin-dir /path/to/buidl-opnet-plugin/buidl"

Replace /path/to/buidl-opnet-plugin with the actual path where you cloned the repo in Step 2.

For example, if you cloned it to your home folder:

alias claudey="claude --plugin-dir ~/buidl-opnet-plugin/buidl"
alias claudeyproj="claude --dangerously-skip-permissions --plugin-dir ~/buidl-opnet-plugin/buidl"

After adding the aliases, reload your shell:

source ~/.bashrc
# or
source ~/.zshrc

What's the difference between the two?

claudey — safe mode. Claude will ask for your approval before running commands. Good for learning or when you want to review each step.

claudeyproj — autonomous mode. Claude runs the whole pipeline without stopping to ask. This is the recommended way once you're comfortable. The /buidl command is designed for this mode — it runs 8 agents back-to-back and stops asking would slow everything down.

#Step 4: Add the opnet-bob MCP

The MCP gives Claude live access to OPNet blockchain data — contract state, transaction status, token info. Without it, Claude has to guess. With it, Claude knows.

claude mcp add opnet-bob --transport http https://ai.opnet.org/mcp

That's it. One command. The MCP is now connected.

#Step 5: Install OPWallet

OPWallet is the Bitcoin wallet that talks to OPNet contracts. Your users will need it to interact with the apps you build, and you'll need it to test everything during development.

Install the OPWallet browser extension from the official OPNet site.

Set it up and create a new wallet. Save your seed phrase somewhere safe.

#Step 6: Get Test Bitcoin

You need test BTC to deploy contracts and run transactions on testnet.

Go to faucet.opnet.org and paste your testnet wallet address. You'll get enough test BTC to build and deploy multiple apps.

#Test Your Setup

Create a new folder for your first project:

mkdir my-first-opnet-app
cd my-first-opnet-app

Launch Claude in autonomous mode:

claudeyproj

If Claude opens and you see the OPNet agents loaded in the plugin info, you're set up correctly.

Try running:

/buidl-status

You should see the pipeline state. Empty for now, which is correct.

#What You Just Installed

Here's what the plugin gives you:

8 specialized agents:

• opnet-contract-dev — writes AssemblyScript smart contracts
• opnet-frontend-dev — builds React frontends with OPNet wallet integration
• opnet-backend-dev — builds backend services and APIs
• opnet-auditor — reviews code against 27 real-bug patterns from production OPNet repos
• opnet-deployer — handles contract deployment to testnet and mainnet
• opnet-e2e-tester — writes and runs end-to-end tests
• opnet-ui-tester — tests the frontend interface
• loop-reviewer — reviews the whole build before marking it complete

27 audit patterns — 9 marked CRITICAL — pulled from real bugs found in the btc-vision repositories. The auditor agent checks your code against all of them automatically.

The /buidl command — runs the full pipeline: idea → challenge → spec → build → review → PR. One command, 8 agents, working together.

You're ready. Next chapter: how to actually talk to Claude and get things built.