Key Features at a Glance
LLM-agnostic, wallet-agnostic, chain-agnostic, modular.
PILSO OS delivers secure, modular infrastructure to let LLM agents safely execute onchain actions using any Web3 wallet.
Here’s what sets it apart:
Wallet-as-Signer Security
Private keys never leave the wallet
All transactions are signed client-side via MetaMask (WalletConnect and others coming soon)
Removes need for custodial agents, bots, or key delegation
LLM-Compatible Execution Layer
Compatible with Claude, GPT-4, DeepSeek and other LLMs
Agents interpret natural language instructions and convert them into secure blockchain operations
Role-based agent personalities ensure scoped behavior (e.g. “signer”, “developer”, “auditor”)
Natural Language Interface
Abstracts away RPCs, ABI formats, and tooling jargon
Prompts like “Deploy this contract on Sepolia Testnet or bridge 50 USDC to Optimism” are interpreted, executed, and signed securely
Enables true agent UX across wallets and chains
Modular MCP Server Framework
Each blockchain task runs on a dedicated MCP server
Server modules are composable and trustless
Examples include:
metamask-mcp– transaction signingchainlist-mcp– chain info fetchsolc-mcp– smart contract compileerc20-mcp– token actions
CLI Runtime for Power Users
Run local agent sessions via terminal
Configure models, roles, environments
Resume, replay, and audit sessions
Used by devs building custom automations or prototypes
Non-Custodial by Design
No token approvals needed
No middleman wallets
No signature hijacking
You sign what you see. Agents execute what you allow.
Extensible for Builders
Add custom MCP servers to integrate with dApps, data sources, or off-chain APIs
Define agent roles with personality, task scope, and execution behavior
Build on top — don’t reinvent the stack
Last updated