The stepping stone to owning your agent goldmine

Your agent works locally. Gem it.

Build it once. Own it. Ship it anywhere.

Most of your best agent work never leaves your terminal — your most valuable, least-captured asset. You've already built and refined real capability with Claude, Codex, or Hermes: your agent setup, your session transcripts, your usage data. AgentGem surfaces that goldmine and distills it into a secret-safe, composable Gem that's yours to own — reuse it, deploy it on demand, and sell it per call when the network opens. Built on AgentBack.

Get started → Browse the marketplace How it works

$ npx @ninemind/agentgem then open localhost:4317

Download the desktop app

for macOS, Windows & Linux

Watch · 60 seconds

See a Gem come to life.

Four steps

You're already sitting on a goldmine.

Surface it AgentGem reads what you've already refined with Claude, Codex, or Hermes — your agent setup, session transcripts, and usage data

Reuse it Analyze distills the procedures you repeat into draft skills — pack them into a secret-safe Gem and run it again in a single call

Deploy it Ship the same Gem anywhere — Claude, OpenAI, and more. It isn't locked to one runtime

Sell it soon Others run it on demand — and you get paid per call. See the vision →

Why a Gem, not a markdown file

A skill is a copy. A Gem is a service.

Today you post a SKILL.md to a registry like skills.sh and promote it on X. People copy it, fork it, or scroll past — and you earn nothing. You can't profit from a markdown file. A Gem is different: a live service others call, not a file they take.

Today — sharing markdown

Post a SKILL.md to a registry, promote it on X.
Anyone copies, forks, or scrolls past it.
You collect stars — and earn nothing.

Gem it

Your agent ships as a live, callable service — secrets sealed.
It runs on demand, discovered agent-to-agent over A2A.
You get paid per call.

Why AgentGem

One contract, every boundary.

One Zod contract

Every operation is defined once. AgentBack exposes it as a REST endpoint, an MCP tool, and an OpenAPI 3.1 doc — with runtime validation for free.

RESTMCPOpenAPI

Redacted at capture

MCP secrets are stripped the moment config is read — by value and by key name. No response, preview, or Gem ever carries a raw secret.

trust boundary

Composable registry

Publish Gems to a GitHub-backed registry, then resolve, merge, and install them — assembling bigger agents out of smaller, shareable pieces.

publishmergeinstall

Deploy anywhere

The same neutral Gem feeds multiple code-gen targets and managed backends — pick a target, publish, and undeploy when you're done.

Eve Flue OpenAI AgentCore Claude Agents

What you get

A Gem, with secrets sealed.

gem.json

// built from your agent config — secrets never leave your machine
{
  "name": "research-assistant",
  "skills": ["deep-research", "make-pdf"],
  "mcpServers": {
    "firecrawl": {
      "command": "npx",
      "env": { "FIRECRAWL_API_KEY": "<redacted>" }
    }
  },
  "claudeMd": "included"
}

The north star

A Shopify for AI agents.

Shopify let anyone launch a store without building commerce infrastructure. AgentGem is laying the same foundation for agents — package once, then publish, deploy, discover, and monetize. Not a catalog of downloads: a live agent service network where Gems run on demand, are discovered agent-to-agent, and get paid per call. The Gem is the stepping stone.

1Package

Your agent project becomes a secret-safe, composable Gem.

Shipped

2Publish

Push Gems to the agent service network for others to find and call.

Shipped

3Deploy on demand

Spin up a Gem as a live service when it's called.

Emerging

4Discover

Agents find and call your services over A2A.

Vision

5Monetize

Charge per call with agent-native payments — x402 / MPP.

Vision

Read the vision →