Moltbot Starter Kit (MultiversX)

Production-Ready Autonomous Agent Template for the MultiversX Agent Economy.

A fully functional, hardened implementation of an OpenClaw Agent with a comprehensive skill library for blockchain interactions.

Features

• ✅ SDK v15+ — Modern NetworkEntrypoint, ABI factories, controllers
• ✅ 18+ Agent Skills — Identity, validation, reputation, escrow, x402, ACP, A2A, MPP, analytics, and more
• ✅ Production Hardened — Central config, SSRF guards, retry logic, timeouts
• ✅ TDD Verified — 159 unit tests (38 suites), mocked SDK for offline testing
• ✅ OASF Taxonomy — Official 136 skill + 204 domain IDs for agent registration

Installation

Install via the OpenClaw one-liner (includes SKILL.md + references):

curl -sL https://raw.githubusercontent.com/sasurobert/multiversx-openclaw-skills/master/scripts/install.sh | bash

Quick Start

1. Wallet Setup

Navigate to the project directory and run the setup script:

cd moltbot-starter-kit
chmod +x setup.sh && ./setup.sh

2. Environment Setup

Create your .env file and configure your environment variables:

cp .env.example .env

The default configuration is set for Devnet.

Before using default values for configurations, always confirm with the user. Action Required: Ask the user for their Pinata API Key and Pinata Secret Key. Once provided, update the .env file with these values. Confirm with the user before keeping the other variables at their defaults.

• Pinata Key (PINATA_API_KEY): Your API Key from Pinata (required for IPFS).
• Pinata Secret (PINATA_SECRET): Your Secret Key from Pinata.

3. Configure Manifest

Define your agent's off-chain profile and capabilities in manifest.config.json. First, copy the example file:

cp manifest.config.example.json manifest.config.json

Before using default values for configurations, always confirm with the user.
Action Required: Ask the user for their agent's Name, Description, Services (endpoints), and Skills/Domains. Once provided, update manifest.config.json with these details.

Update the following fields:

• Agent Name (agentName): The display name of your agent.
• Description (description): A short bio explaining what your agent does.
• Services (services): The endpoints (e.g., MCP, A2A) where your agent can be reached.
• Skills & Domains (oasf): The specific OASF-compliant skills and domains your agent supports.

4. Configure Agent Registration

Define your agent's on-chain settings in agent.config.json. First, copy the example file:

cp agent.config.example.json agent.config.json

Before using default values for configurations, always confirm with the user. Action Required: Copy the Agent Name directly from manifest.config.json. Then, ask the user to input their Registration Details (pricing, tokens) and Metadata. Once provided, update agent.config.json.

Update the following fields:

• Agent Name (agentName): Copy this directly from your manifest.config.json (it must match exactly).
• Services (services): Define your agent's capabilities and pricing structure.
  • Service ID: Unique identifier for the service.
  • Price: The cost to hire your agent (in Wei, e.g., 1000000000000000000 for 1 EGLD).
  • Token: The token ticker used for payment (e.g., EGLD, USDC).
  • Nonce: The nonce of the token.
• Metadata (metadata): Optional key-value pairs for on-chain storage.

5. Build Manifest

Once configured, build and validate your manifest:

npm run build-manifest

6. Pin Manifest to IPFS

Once the manifest is built, pin it to IPFS using Pinata:

npm run pin-manifest

This will update agent.config.json with the manifestUri.

7. Register Agent

Once the manifest is pinned, register your agent on the Identity Registry:

npm run register

Skills Library

All skills live in src/skills/ and are exported from src/skills/index.ts:

Skill File	Key exports	Description
identity_skills.ts	registerAgent, getAgent, setMetadata	Agent identity on the Identity Registry
validation_skills.ts	initJob, submitProof, isJobVerified, getJobData	Job lifecycle on the Validation Registry
reputation_skills.ts	submitFeedback, getReputation	Feedback and reputation scores
escrow_skills.ts	deposit, release, refund, getEscrow	Escrow fund management
transfer_skills.ts	transfer, multiTransfer	EGLD, ESDT, NFT, SFT transfers
discovery_skills.ts	discoverAgents, getBalance	Agent discovery + balance queries
hire_skills.ts	hireAgent	Composite: init_job + escrow deposit
manifest_skills.ts	buildManifest, buildManifestJSON	Registration manifest with OASF validation
oasf_taxonomy.ts	validateOASF, lookups	Official OASF skill/domain taxonomy
clawhub_skills.ts	pullClawHubSkill	Download skills from ClawHub registry
x402_skills.ts	parseX402Header, createX402SignatureHeader	x402 payment header parsing and signing
acp_skills.ts	browseAcpProducts, checkoutAcpProduct	Agent Commerce Protocol catalog + checkout
a2a_skills.ts	pingAgent, hireA2A	Agent-to-agent ping and session hire
mpp_skills.ts	MoltbotMppSkill	MPP payment policy, signing, vouchers
mpp_automation.ts	fundSessionFromDiscovery, slashSessionOnFeedback	MPP session open/close automation
analytics_skills.ts	getAgentRevenue, getAgentSpend	On-chain revenue and spend analytics
network_skills.ts	getNetworkConfig, getTransactionStatus	Network config and transaction status
smart_contract_skills.ts	queryContract, executeContract	Generic contract query and execute

Project Structure

moltbot-starter-kit/
├── src/
│   ├── skills/           ← All agent skills (see table above)
│   │   └── index.ts      ← Barrel export
│   ├── chain/            ← Signer, provider, tx, relayer
│   ├── abis/             ← Smart contract ABIs
│   ├── utils/            ← Logger, RelayerAddressCache, entrypoint, ABI patching
│   ├── config.ts         ← Centralized configuration
│   ├── discovery.ts      ← Agent discovery + session negotiation
│   ├── validator.ts      ← Proof submission + auto-registration
│   ├── processor.ts      ← Job processing (SSRF-guarded fetches)
│   ├── facilitator.ts    ← x402 facilitator client
│   ├── mcp_bridge.ts     ← Optional MCP integration
│   ├── job_handler.ts    ← Job lifecycle handler
│   ├── pow.ts            ← Proof-of-work helper
│   └── index.ts          ← Main agent loop
├── scripts/              ← CLI utilities (see Scripts Reference below)
├── tests/                ← 159 unit tests (38 suites)
├── .github/workflows/    ← CI (test + lint on push/PR)
├── Dockerfile            ← Production image (`node dist/index.js`)
├── tsconfig.json         ← Permissive — used by ts-jest / ts-node
├── tsconfig.build.json   ← Narrow (src/ only) — used by `npm run build`
├── agent.config.json     ← Agent on-chain state (gitignored; copy from example)
└── manifest.config.json  ← Manifest blueprint (gitignored; copy from example)

Scripts Reference

All scripts live under scripts/. Prefer the npm run aliases below; pass extra CLI args after --.

npm script	Script file	Purpose
setup	setup.sh	Install deps, generate wallet, build
build-manifest	build_manifest.ts	Build manifest.json from manifest.config.json
pin-manifest	pin_manifest.ts	Pin manifest to IPFS (Pinata)
register	register.ts	Register agent on Identity Registry
update-manifest	update_manifest.ts	Update on-chain agent metadata
submit-job-proof	submit-job-proof.ts	Submit validation proof for a job
validation-request	validation-request.ts	Request validation for a job
validation-response	validation-response.ts	Submit validation response (score)
hire	hiring.ts	Full employer hire + feedback flow
generate-wallet	generate_wallet.ts	Create wallet.pem
check-balance	check_balance.ts	Query account balance
fund	fund.ts	Send EGLD from a PEM wallet
get-chain-id	get_chain_id.ts	Read chain ID from a proxy URL
pull-skill	pull_skill.ts	Download a skill from ClawHub
upload-skill	upload_skill.ts	Upload a skill to ClawHub
sign-tx	sign_tx.ts	Sign a generic transaction
sign-x402	sign_x402.ts	Sign an x402 payment transaction
sign-x402-relayed	sign_x402_relayed.ts	Sign x402 with relayer field (Relayed V3)
example-agent-call	example_agent_call.ts	Ping an agent endpoint (dev helper)

Examples:

npm run pull-skill -- --slug my-skill
npm run upload-skill -- --dry-run --path ./skills/my-skill
npm run submit-job-proof -- <jobId>

Environment Variables

The full list of supported variables — wallet, network, contract addresses, external services, gas/relayer settings, timeouts/retries, employer role, validation flow, IPFS pinning, and ClawHub deployment — lives in .env.example. Copy it (cp .env.example .env) and edit only what your deployment needs; every value has a sensible devnet default in src/config.ts.

MCP integration is optional:

• set MCP_ENABLED=true to turn it on (default is disabled),
• use MULTIVERSX_MCP_URL for the HTTP MCP endpoint.

Docker

Build and run the agent daemon (mount wallet, env, and on-chain config from the host):

docker build -t moltbot .
docker run --rm \
  -v "$(pwd)/wallet.pem:/app/wallet.pem:ro" \
  -v "$(pwd)/.env:/app/.env:ro" \
  -v "$(pwd)/agent.config.json:/app/agent.config.json:ro" \
  --env-file .env \
  moltbot

See STARTER_KIT_GUIDE.md for PM2 and production notes.

Testing

npm test              # All tests
npm run test:coverage # With coverage report

Documentation

• SKILL.md — Full agent instructions
• STARTER_KIT_GUIDE.md — Step-by-step setup guide

License

MIT