HN Showcase
Back to browse
SmartAgentKit – policy-governed smart wallets for AI agents

SmartAgentKit – policy-governed smart wallets for AI agents

by martinbf·Mar 4, 2026·2 points·0 comments
Visit ProjectView on HN

AI Analysis

●●SolidBig BrainSolve My Problem

Policy-enforced wallets for agents, but ERC-4337 smart accounts already commoditized.

Strengths
  • •Session key + spending limit combo genuinely constrains agent autonomy on-chain.
  • •Modular policy hooks (Safe + ERC-7579) let developers build custom guardrails.
  • •LangChain integration makes agent integration immediate; mock testing avoids deployment friction.
Weaknesses
  • •Safe smart accounts + Pimlico are well-established; core insight (policies for agents) is the only novelty.
  • •Blockchain-first approach adds deployment cost and latency vs in-memory constraints.
Category
Finance
Target Audience

AI agent developers building blockchain-interacting systems; Web3 infrastructure teams.

Similar To

Safe smart accounts · MakerDAO DSProxy guards · Gnosis Safe policies

Post Description

I built SmartAgentKit to explore a problem that is becoming more common as AI agents interact with crypto systems.

Most agent implementations simply give the agent a wallet key. That works, but it provides no constraints on what the agent can do.

SmartAgentKit introduces policy-governed wallets using ERC-4337 smart accounts and modular policy hooks.

Policies are enforced on-chain and can include things like: - spending limits - contract allowlists - emergency pause - session-based access control

The system builds on: - Safe smart accounts - ERC-7579 modular account architecture - Rhinestone ModuleKit / Smart Sessions - Pimlico bundler + permissionless.js

Developers can also deploy custom policy contracts and install them in wallets.

Docs: https://smartagentkit.xyz

Repo: https://github.com/smartagentkit/smartagentkit

Similar Projects