Snak
An agent engine for creating powerful and secure AI Agents powered by Starknet.
A Agent Engine for creating powerful and secure AI Agents powered by Starknet. Available as both an NPM package and a ready-to-use backend.
Quick Start
Prerequisites
- Starknet wallet (recommended: Argent X)
- AI provider API key (Anthropic/OpenAI/Google Gemini/Ollama)
- Docker and Docker compose installed
- Node.js and pnpm installed
Installation
git clone https://github.com/kasarlabs/snak.git
cd snak
pnpm install
Configuration
- Create a
.envfile by copying.env.example:
cp .env.example .env
Then, fill in the necessary values in your .env file:
# --- Starknet configuration (mandatory) ---
STARKNET_PUBLIC_ADDRESS="YOUR_STARKNET_PUBLIC_ADDRESS"
STARKNET_PRIVATE_KEY="YOUR_STARKNET_PRIVATE_KEY"
STARKNET_RPC_URL="YOUR_STARKNET_RPC_URL"
# --- AI Model API Keys (mandatory) ---
# Add the API keys for the specific AI providers you use in config/models/default.models.json
# The agent will automatically load the correct key based on the provider name.
# Example for OpenAI:
OPENAI_API_KEY="YOUR_OPENAI_API_KEY" # (e.g., sk-...)
# Example for Anthropic:
ANTHROPIC_API_KEY="YOUR_ANTHROPIC_API_KEY" # (e.g., sk-ant-...)
# Example for Google Gemini:
GEMINI_API_KEY="YOUR_GEMINI_API_KEY"
# Example for DeepSeek:
DEEPSEEK_API_KEY="YOUR_DEEPSEEK_API_KEY"
# Note: You do not need an API key if using a local Ollama model.
# --- General Agent Configuration (mandatory) ---
SERVER_API_KEY="YOUR_SERVER_API_KEY" # A secret key for your agent server API
SERVER_PORT="3001"
# --- PostgreSQL Database Configuration (mandatory) ---
POSTGRES_USER=admin
POSTGRES_HOST=localhost
POSTGRES_DB=postgres
POSTGRES_PASSWORD=admin
POSTGRES_PORT=5432
# --- LangSmith Tracing (Optional) ---
# Set LANGSMITH_TRACING=true to enable tracing
LANGSMITH_TRACING=false
LANGSMITH_ENDPOINT="https://api.smith.langchain.com"
LANGSMITH_API_KEY="YOUR_LANGSMITH_API_KEY" # (Only needed if LANGSMITH_TRACING=true)
LANGSMITH_PROJECT="Snak" # (Optional project name for LangSmith)
# --- Node Environment ---
NODE_ENV="development" # "development" or "production"
-
Configure AI Models (Optional): The
config/models/default.models.jsonfile defines the default AI models used for different tasks (fast,smart,cheap). You can customize this file or create new model configurations (e.g.,my_models.json) and specify them when running the agent. Seeconfig/models/example.models.jsonfor the structure.The agent uses the
providerfield in the model configuration to determine which API key to load from the.envfile (e.g., ifproviderisopenai, it loadsOPENAI_API_KEY). -
Create your agent configuration file (e.g.,
default.agent.jsonormy_agent.json) in theconfig/agents/directory:
{
"name": "Your Agent name",
"group": "Your Agent group",
"description": "Your AI Agent Description",
"lore": ["Some lore of your AI Agent 1", "Some lore of your AI Agent 1"],
"objectives": [
"first objective that your AI Agent need to follow",
"second objective that your AI Agent need to follow"
],
"knowledge": [
"first knowledge of your AI Agent",
"second knowledge of your AI Agent"
],
"interval": "Your agent interval beetween each transaction of the Agent in ms,",
"chatId": "Your Agent Chat-id for isolating memory",
"max_iterations": "The number of iterations your agent will execute before stopping",
"mode": "The mode of your agent, can be interactive, autonomous or hybrid",
"memory": {
"enabled": "true or false to enable or disable memory",
"shortTermMemorySize": "The number of messages your agent will remember"
},
"plugins": ["Your first plugin", "Your second plugin"],
"mcp_servers": {
"nxp_server_example": {
"command": "npx",
"args": ["-y", "@npm_package_example/npx_server_example"],
"env": {
"API_KEY": "YOUR_API_KEY"
}
},
"local_server_example": {
"command": "node",
"args": ["node /path/to/local_server/dist/index.js"]
}
}
}
You can simply create your own agent configuration using our tool on snakagent
Usage
Prompt Mode
Run the promt:
# start with the default.agent.json
pnpm run start
# start with your custom configuration
pnpm run start --agent="name_of_your_config.json" --models="name_of_your_config.json"
Server Mode
Run the server :
# start with the default.agent.json
pnpm run start:server
# start with your custom configuration
pnpm run start:server --agent="name_of_your_config.json" --models="name_of_your_config.json"
Available Modes
| Interactive Mode | Autonomous Mode | |
|---|---|---|
| Prompt Mode | ✅ | ✅ |
| Server Mode | ✅ | ✅ |
Implement Snak in your project
- Install snak package
#using npm
npm install @snakagent
# using pnpm
pnpm add @snakagent
- Create your agent instance
import { SnakAgent } from 'starknet-agent-kit';
const agent = new SnakAgent({
provider: new RpcProvider({ nodeUrl: process.env.STARKNET_RPC_URL }),
accountPrivateKey: process.env.STARKNET_PRIVATE_KEY,
accountPublicKey: process.env.STARKNET_PUBLIC_ADDRESS,
aiModel: process.env.AI_MODEL,
aiProvider: process.env.AI_PROVIDER,
aiProviderApiKey: process.env.AI_PROVIDER_API_KEY,
signature: 'key',
agentMode: 'interactive',
agentconfig: y,
});
const response = await agent.execute("What's my ETH balance?");
Actions
To learn more about actions you can read this doc section. A comprehensive interface in the Kit will provide an easy-to-navigate catalog of all available plugins and their actions, making discovery and usage simpler.
To add actions to your agent you can easily follow the step-by-steps guide here
Contributing
Contributions are welcome! Feel free to submit a Pull Request.
License
MIT License - see the LICENSE file for details.
For detailed documentation visit docs.kasar.io
相關伺服器
Alpha Vantage MCP Server
贊助Access financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
Chronos Protocol
A robust MCP server that eliminates temporal blindness in AI coding agents through intelligent time tracking, persistent memory, and complete session traceability.
Add Tool
Fetches and sets up an MCP server from a specified GitHub repository.
ucn
Universal Code Navigator - a lightweight MCP server that gives AI agents call-graph-level understanding of code. Instead of reading entire files, agents ask structural questions like: "who calls this function", "what breaks if I change it", "what's unused", and get precise, AST-verified answers. UCN parses JS/TS, Python, Go, Rust, Java, and HTML inline scripts with tree-sitter, then exposes 28 navigation commands as a CLI tool, MCP server, or agent skill.
Maven Package README MCP Server
Search for and retrieve detailed information, including READMEs and metadata, for Maven packages from Maven Central.
ITerm MCP Server
An MCP server for iTerm2, enabling AI assistants to interact with the terminal.
302AI Basic MCP Server
A toolkit for enhancing the fundamental capabilities of large language models.
Remote MCP Server on Cloudflare
A customizable remote MCP server for deployment on Cloudflare Workers, operating without authentication.
Atomic APIs
17 developer utility APIs as MCP tools — PII redaction, prompt injection detection, web-to-markdown, WCAG scanning, receipt OCR, and more — zero configuration, sub-second responses.
MCP Proxy
A proxy server for MCP requests, supporting SSE and stdio transports.
PHP MCP Server
A server-side implementation of the Model Context Protocol (MCP) for PHP applications, allowing exposure of application parts as standardized MCP Tools, Resources, and Prompts.