Main HubSupport
Guides
Start typing to search…

Learn to Code with Ai

Learn how to code with AI. Step through interactive tutorials that turn plain-language instructions into working Solana code. Carbium’s AI-ready documentation helps you move from idea to deployment through guided, reproducible examples.

Communicating with AI

Before writing code, it’s essential to understand how to talk to an AI assistant effectively. AI models understand structured language, clarity and context improve their accuracy.

Best practices:

  • Be explicit: specify which language (Rust, TypeScript, Python, etc.) or framework you’re using.
    • Python is easy-to-approach and powerful for DeFi activities.
  • Include details about the environment (Solana SDK, CLI tools, RPC endpoint).
  • Request step-by-step output or ask the AI to explain what each part of the code does.
  • Iterate: if a response is incomplete, refine your prompt rather than restarting.

Example prompt:

“Write a Python script that trades SOL for USDC using the Carbium DEX API endpoint and logs transaction signatures.”

Supported AI Assistants

You can use any LLM interface capable of structured code generation. Every major artifical intelligence does have Code intepreter features:

  • OpenAI GPT models (GPT-5, Especially Codex)
  • Anthropic Claude
  • Perplexity, Gemini, or local LLMs

To get the best results, share the Carbium documentation link with your assistant, otherwise you can share snippets from our documentation:

https://docs.carbium.io

This allows the model to reference Carbium’s official endpoints, syntax, and RPC structure.

Integrating AI via CLI or Editor

You can directly integrate GPT into your coding workflow using the OpenAI CLI or Visual Studio Code extensions.

Option A – Using the OpenAI CLI

You’ll need an OpenAI API key and the openai command-line interface installed.

Installation

pip install openai

Authenticate

export OPENAI_API_KEY="your_api_key_here"

Example Usage

openai api.chat_completions.create \
  -m gpt-4-turbo \
  -g "Generate a Rust function that initializes a Solana account with Carbium RPC"

See the official OpenAI reference for all CLI flags: ➡ OpenAI CLI documentation

Option B – Using Visual Studio Code

  1. Install the ChatGPT Codex CLI or Claude Code integration
  2. Add your API key in the settings.
  3. You can then query GPT directly in the sidebar or inline editor.
  4. Paste snippets from Carbium docs

Option C – Using Apple Terminal with GPT or Claude

If you prefer working natively on macOS, you can run both OpenAI GPT and Anthropic Claude models directly from the Apple Terminal. This approach is ideal for quick code generation, prototyping, or integrating Carbium’s SDK commands without switching editors.

1. Set Up the Environment

Ensure that Python 3 and pip are installed on your system. Then install the official OpenAI and Anthropic Python clients:

pip install openai anthropic

Set your API keys as environment variables:

export OPENAI_API_KEY="your_openai_api_key"
export ANTHROPIC_API_KEY="your_anthropic_api_key"

(You can add these lines to ~/.zshrc or ~/.bash_profile to make them permanent.)

2. Run GPT from the Terminal

Generate code directly through the OpenAI Python client:

python3 - <<'PY'
from openai import OpenAI

client = OpenAI()
prompt = "Generate a Python script that queries Carbium RPC and prints the latest Solana slot number."

resp = client.chat.completions.create(
    model="gpt-4-turbo",
    messages=[{"role": "user", "content": prompt}]
)

print(resp.choices[0].message.content)
PY

3. Run Claude from the Terminal

Use Anthropic’s client for Claude 3 Opus or Claude Instant:

python3 - <<'PY'
from anthropic import Anthropic

client = Anthropic()
prompt = "Write a Rust example that initializes a Solana wallet using Carbium SDK."

resp = client.messages.create(
    model="claude-3-opus-20240229",
    max_tokens=500,
    messages=[{"role": "user", "content": prompt}]
)

print(resp.content[0].text)
PY

4. Combine with Carbium Docs

Reference the Carbium documentation directly in your prompt to ensure the AI uses official APIs and syntax:

Reference docs: https://docs.carbium.io
Task: Generate a Solana transaction using Carbium’s DEX API.

This improves accuracy and guarantees compatibility with Carbium’s RPC and SDK layers.

Technical References

5. Next Steps

Once you’re comfortable communicating with AI, continue to:

Updated about 7 hours ago

What’s Next