This guide is written for openclaw users who want to ship a live, paid AI agent using aixyz. You do not need prior coding experience or knowledge of crypto infrastructure — every step is explained from scratch.
Works with
AI-first code editor with Composer
Before installing skills in Cursor, ensure your development environment meets these requirements:
node --versionaixyz-on-openclawExecute the skills CLI command in your project's root directory to begin installation:
Fetches aixyz-on-openclaw from agentlyhq/aixyz and configures it for Cursor.
The CLI shows a list of agents. Use arrow keys and space to select Cursor:
Confirm successful installation by checking the skill directory location:
Restart Cursor to activate aixyz-on-openclaw. Access via /aixyz-on-openclaw in your agent's command palette.
We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.
Skills execute code in your environment. Always review source, verify the publisher, and test in isolation before production.
Submit your Claude Code skill and start earning
Automate repetitive workflows and reduce manual effort
Example
Generate reports, summarize documents, draft communications
Save 3-5 hours per week on routine tasks
Learn new skills, understand complex topics, get expert guidance
Example
Explain concepts, provide examples, suggest learning resources
Accelerate learning and skill development by 2x
Enhance output quality through reviews, suggestions, and refinements
Example
Review drafts, suggest improvements, catch errors
Improve work quality by 30-40% with less effort
0
total installs
0
this week
81
GitHub stars
0
upvotes
Run in your terminal
0
installs
0
this week
81
stars
This guide is written for openclaw users who want to ship a live, paid AI agent using aixyz. You do not need prior coding experience or knowledge of crypto infrastructure — every step is explained from scratch.
Quick reference: The general aixyz skill lives at skills.sh/agentlyhq/aixyz. Consult it for full CLI flags and advanced topics. This guide focuses on the path that matters most to openclaw users.
An AI agent that:
aixyz requires Bun (not Node). If you already have Bun ≥ 1.3, skip this step.
# macOS / Linux
curl -fsSL https://bun.sh/install | bash
# Windows (PowerShell, requires WSL or Windows 10+)
powershell -c "irm bun.sh/install.ps1 | iex"
Verify:
bun --version # should print 1.3 or higher
Why Bun? The aixyz build pipeline uses Bun's native build API and plugin system — this is a hard architectural requirement, not a preference. The CLI (
@aixyz/cli) cannot run on Node. Bun also includes a built-in test runner, which means no extra test framework is needed. Curious about the full rationale? Read Why Bun.
CRITICAL: You MUST use
create-aixyz-appto create new agent projects. Do NOT manually createaixyz.config.ts,package.json,app/agent.ts, or any other project files by hand. Manually creating these files WILL result in broken builds, missing dependencies, and incorrect configurations. Always scaffold withbunx create-aixyz-appfirst, then modify the generated files.
# Create a new agent project (uses all defaults)
bunx create-aixyz-app my-agent --yes
cd my-agent
This creates:
my-agent/
aixyz.config.ts ← agent name, description, payment config
app/
agent.ts ← your AI agent logic
tools/ ← tool files (optional)
package.json
.env.local ← API keys (never commit this file)
Open .env.local and add your LLM provider's API key. The default scaffold uses @ai-sdk/openai, so add:
OPENAI_API_KEY=sk-...
If you prefer a different provider (Anthropic, Google, Amazon Bedrock, etc.), swap the @ai-sdk/* adapter in package.json and app/agent.ts, then set the corresponding env var instead. See ai-sdk.dev for the full list of providers.
Run locally to test:
bun run dev
# Agent is now live at http://localhost:3000
Testing with an AI agent client? Install the
use-agentlyskill so your AI agent can call your local agent in dev mode:npx skills add https://github.com/agentlyhq/use-agently --skill use-agentlyThis skill lets any AI agent discover and call your locally running agent at
http://localhost:3000. While testing locally, set"scheme": "free"so callers are not charged (see Step 7). Before going to production, switch to"scheme": "exact"so your agent earns from every request.
Your agent needs a public HTTPS URL so other agents and clients can reach it. If you already have a preferred hosting platform, use it. If you are starting fresh, here are our recommendations.
Vercel gives you a free HTTPS URL with zero configuration. It is the easiest path.
bun run build and the Output Directory to .vercel/outputOPENAI_API_KEY) in the Vercel dashboard under Settings → Environment Variableshttps://my-agent.vercel.appVercel auto-deploys on every push. Your agent is serverless and scales automatically.
Use these if your agent needs to hold state between requests or run background jobs.
| Platform | Free tier | Notes |
|---|---|---|
| Railway | 5 USD/month credit | Easiest Docker-free deploy |
| Render | Generous free tier | Sleeps after inactivity |
| Fly.io | Free allowance | More control, steeper learning curve |
For all three: push your repo, connect the platform, set your LLM provider API key (e.g., OPENAI_API_KEY) as an environment variable, and follow their deploy wizard. They all detect Bun automatically.
If you are already running OpenClaw on a cloud platform (AWS, GCP, Azure, DigitalOcean, etc.), the simplest path is to expose the agent through that same platform:
🔒 Least-privilege rule: When exposing the agent process, only open the port the agent listens on (default
3000). Do not grant the process broader network, IAM, or filesystem access than it needs to serve HTTP requests.
If you just need a quick tunnel for local testing, use ngrok:
# ngrok — creates a public HTTPS URL that forwards to localhost:3000
npx ngrok http 3000
⚠️ Security warning: Local tunnels are fine for short-lived testing but should never be used for a production agent. Anyone who discovers the URL can send requests to your machine. Always use a proper hosting provider for live traffic.
You need a wallet to:
use-agently (simplest, recommended for openclaw users)use-agently is the simplest wallet for the aixyz ecosystem. It is designed for the circular economy of agents paying agents.
When prompted during aixyz erc-8004 register, select "Generate a new wallet" and the CLI will create one for you and display the private key.
🔐 Private key security — read this carefully:
- Never share your private key with anyone. Anyone who has it controls your funds.
- Never commit it to git or paste it in public channels (Discord, Slack, GitHub, etc.).
- Write it down offline and store it somewhere safe (e.g., a password manager).
- For a production agent that earns significant income, consider a hardware wallet (Ledger, Trezor).
0x…)Any wallet that supports Ethereum-compatible networks works: Coinbase Wallet, Rabby, Frame, etc.
Your wallet address is your "bank account number." You share it publicly. Your private key / seed phrase is your password — never share it, never commit it to code.
Registering on ERC-8004 costs a small gas fee (a fraction of a dollar). You need the native coin of your chosen network.
| Network | Coin needed | Where to get it |
|---|---|---|
| Ethereum mainnet | ETH | Coinbase, Binance, Kraken |
| BNB Smart Chain | BNB | Binance |
| Base | ETH | Coinbase (Base is a Coinbase L2) |
| Polygon | POL | Coinbase, Binance |
Recommended network: Base. The x402 payment facilitator is primarily deployed on Base, so your agent will receive payments most reliably there. Base also has very low gas fees — registration typically costs a few cents. Ethereum mainnet gas varies widely and is usually not worth it for getting started. Gas prices on all networks can fluctuate; the amounts above are rough guides only.
Steps to fund your wallet:
No crypto yet? You can still build and run your agent locally and on Vercel without registration. Register once you are ready to be publicly discoverable.
ERC-8004 is an on-chain registry that makes your agent discoverable by other agents, platforms, and users. It is like a DNS entry but for AI agents.
Once your agent is deployed (Step 3) and you have a funded wallet (Steps 4–5), run:
# Interactive mode (will prompt you for each setting)
aixyz erc-8004 register
# Non-interactive mode (all values as flags)
aixyz erc-8004 register \
--url https://my-agent.vercel.app \
--chain base \
--broadcast
The CLI will ask for (or accept as flags):
| Prompt | Flag | What to enter |
|---|---|---|
| Agent URL | --url |
Your deployed URL from Step 3 |
| Chain | --chain |
base (recommended), ethereum, bsc, polygon |
| Wallet private key | --private-key |
The key from your wallet (never share this) |
| Trust mechanisms | --trust |
Accept the default |
| Broadcast | --broadcast |
Pass this flag to actually submit on-chain |
See all flags:
aixyz erc-8004 register --help
After registration, your agent appears in the ERC-8004 registry and becomes discoverable by the ecosystem.
Add an accepts export to app/agent.ts so callers are charged per request:
import type { Accepts } from "aixyz/accepts";
// During local testing — callers are not charged
export const accepts: Accepts = { scheme: "free" };
// Production — callers pay $0.005 per request
// export const accepts: Accepts = { scheme: "exact", price: "$0.005" };
Testing tip: Use
"scheme": "free"while developing and testing locally (install theuse-agentlyskill to call your local agent from an AI client). Before going to production, switch to"scheme": "exact"with apriceso your agent earns from every request.
Set the wallet address that receives payments in aixyz.config.ts:
export default {
name: "my-agent",
// ...
x402: {
payTo: "0xYourWalletAddress", // ← paste your address from Step 4
},
} satisfies AixyzConfig;
Or pass it at scaffold time:
bunx create-aixyz-app my-agent --yes --pay-to 0xYourWalletAddress
Payments flow directly to your wallet — no platform takes a cut.
Before announcing your agent, verify:
bun run dev works locally and the agent responds correctlyOPENAI_API_KEY) and any other secrets are set as environment variables on the platform, not committed to your repoaixyz.config.ts has the correct name, description, and payTo addressaccepts export is set so your agent earns from requestsRegistration is only half the work. Getting on-chain doesn't automatically bring users. You need to actively promote your agent.
description in aixyz.config.ts — this appears in the A2A agent card and on-chain registryapp/icon.png — a good icon makes your agent stand out in listings/.well-known/agent-card.json URL — it is the standard entry point for agent-to-agent discovery$0.001–$0.01 per request and adjust based on demandaixyz erc-8004 update --helpRestart your terminal after installing Bun, or run source ~/.bashrc (Linux) / source ~/.zshrc (macOS).
Add your LLM provider's API key to .env.local (local dev) or to your platform's environment variable settings (production).
Your wallet does not have enough gas. Add a few more dollars of the relevant coin and retry.
Check that your platform's build command is bun run build and that all environment variables are set. Check the platform's deploy logs for errors.
Verify your payTo address in aixyz.config.ts is correct and that you rebuilt and redeployed after changing it.
Prerequisites
Time Estimate
15-45 minutes depending on use case complexity
Steps
Common Pitfalls
✓ Do
✗ Don't
💡 Pro Tips
✓ Use when
Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.
✗ Avoid when
Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.
davila7/claude-code-templates
intellectronica/agent-skills
am-will/codex-skills
sickn33/antigravity-awesome-skills
myzy-ai/dokie-ai-ppt
sickn33/antigravity-awesome-skills
aixyz-on-openclaw is among the better-maintained entries we tried; worth keeping pinned for repeat workflows.
Solid pick for teams standardizing on skills: aixyz-on-openclaw is focused, and the summary matches what you get after install.
Useful defaults in aixyz-on-openclaw — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.
Registry listing for aixyz-on-openclaw matched our evaluation — installs cleanly and behaves as described in the markdown.
aixyz-on-openclaw fits our agent workflows well — practical, well scoped, and easy to wire into existing repos.
Keeps context tight: aixyz-on-openclaw is the kind of skill you can hand to a new teammate without a long onboarding doc.
I recommend aixyz-on-openclaw for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.
We added aixyz-on-openclaw from the explainx registry; install was straightforward and the SKILL.md answered most questions upfront.
Solid pick for teams standardizing on skills: aixyz-on-openclaw is focused, and the summary matches what you get after install.
aixyz-on-openclaw is among the better-maintained entries we tried; worth keeping pinned for repeat workflows.
showing 1-10 of 39