npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@usecarry/agents

v0.1.0

Published

Proof-carrying memory for any agent framework — gate memory before generation, get an Answer Receipt. Adapters for LangGraph, the OpenAI Agents SDK, and anything else.

Readme

@usecarry/agents

Proof-carrying memory for any agent framework.

An agent should only read the memory it was granted, and you should be able to prove which memory an answer used. This package gates memory before generation — a blocked namespace is never fetched, rather than fetched and filtered — and hands back a receipt describing what the model could actually see.

Nothing here imports a framework, so adding Carry never drags in a peer dependency.

npm i @usecarry/agents

Hosted Carry

Point it at a gateway and the gate runs server-side against the policy on Sui.

import { createGatewayStore } from "@usecarry/agents";

const store = createGatewayStore({
  baseUrl: "https://usecarry.xyz",
  apiKey: process.env.CARRY_API_KEY!,
});

A failed call returns no memories rather than falling back to ungated recall — an outage must not quietly become an access-control bypass.

LangGraph

import { carryLangGraphNode } from "@usecarry/agents";

graph.addNode("memory", carryLangGraphNode({ store, agent: "support-agent" }));
graph.addEdge("memory", "model");

It returns only the messages to append, which is the update shape LangGraph's message reducers expect, so it drops into an existing graph untouched.

OpenAI Agents SDK

import { carryInstructions } from "@usecarry/agents";

const agent = new Agent({
  name: "support",
  instructions: carryInstructions("You are a support agent.", { store, agent: "support-agent" }),
});

Your base instructions always survive; authorized memory is appended beneath them.

Anything else

Every adapter is a wrapper around one primitive:

import { gateMemory } from "@usecarry/agents";

const { systemPrompt, memories, blockedNamespaces, receipt } = await gateMemory(question, {
  store,
  agent: "support-agent",
  onReceipt: (r) => console.log(r.used, r.blockedNamespaces),
});

systemPrompt is null when nothing was authorized — which is the correct outcome, not an error.

Default-deny

createMemoryStore (for local development and tests) grants nothing until the policy says so, matching the on-chain gate:

const store = createMemoryStore({
  memories,
  policy: { "support-agent": { support: true, billing: true } },
});

An agent nobody configured reads nothing, and a namespace nobody mentioned is not readable. A permissive default is how an unconfigured agent ends up reading everything.

Verify it

Answers anchored through a Carry gateway get a /verify/<id> page that reads the receipt from Sui, re-hashes the Walrus blob and recomputes the verdict — with no wallet, trusting nobody's servers. See usecarry.xyz.

MIT.