GuardMesh CLI

CLI framework for building governed AI agents on 0G with GuardMesh

GuardMesh CLI is a developer tool for creating, testing, and deploying AI agents with built-in governance, TEE-verified decisions, and immutable audit trails on the 0G network.

Features

• 🚀 Quick Start: Initialize agent projects with templates
• 🔐 Policy Management: Define and enforce agent permissions
• 🧪 Testing Framework: Test actions against guardian mesh before deployment
• 📊 Audit Trail: Query and verify decisions on 0G Storage
• 🌐 Guardian Management: Monitor and manage guardian nodes
• 🛠️ Developer-Friendly: Simple CLI commands for complex operations

Installation

npm install -g @guardmesh/cli

Or use directly with npx:

npx @guardmesh/cli init my-agent

Quick Start

1. Initialize a New Agent

guardmesh init my-agent --template code-analyzer
cd my-agent
npm install

2. Configure Environment

cp .env.example .env
# Edit .env with your configuration

3. Configure GuardMesh

guardmesh config --rpc https://evmrpc-testnet.0g.ai
guardmesh config --registry 0xYourRegistryAddress
guardmesh config --audit 0xYourAuditAddress

4. Register Your Agent

guardmesh agent register \
  -i my-agent-01 \
  -r code_analysis_only \
  -a "read_file,code_analysis" \
  -d "write_file,change_permissions"

5. Test Actions

guardmesh test action \
  -a my-agent-01 \
  --action read_file \
  --target auth.py

6. Deploy

guardmesh deploy --config agent.yaml

Commands

guardmesh init

Initialize a new agent project with templates.

guardmesh init [name] [options]

Options:
  -t, --template <type>  Agent template (code-analyzer, data-bot, custom)
  -d, --dir <path>       Target directory

guardmesh agent

Manage agents and policies.

# Register a new agent
guardmesh agent register -i <agentId> -r <role> [-a <allowed>] [-d <denied>]

# List all agents
guardmesh agent list

# Get agent details
guardmesh agent info <agentId>

guardmesh test

Test agent actions against guardians.

# Test a single action
guardmesh test action -a <agentId> --action <type> --target <resource>

# Test policy configuration
guardmesh test policy -a <agentId>

guardmesh deploy

Deploy agent with governance.

guardmesh deploy [options]

Options:
  -c, --config <path>  Agent config file (default: agent.yaml)
  --dry-run            Simulate deployment

guardmesh audit

Query audit trail and decisions.

# Watch live audit feed
guardmesh audit watch [-a <agentId>] [-o <outcome>]

# Query audit history
guardmesh audit history [-a <agentId>] [-l <limit>]

# Verify decision on 0G Storage
guardmesh audit verify <merkleRoot>

guardmesh guardians

Manage guardian nodes.

# Show guardian status
guardmesh guardians status

# Add a guardian
guardmesh guardians add -p <pubkey> -e <ens> [--endpoint <url>]

# Remove a guardian
guardmesh guardians remove <pubkey>

guardmesh config

Configure GuardMesh CLI.

guardmesh config [options]

Options:
  --rpc <url>           Set RPC endpoint
  --registry <address>  Set registry contract address
  --audit <address>     Set audit contract address
  --show                Show current configuration

Agent Templates

Code Analyzer

An agent restricted to code analysis tasks.

guardmesh init my-analyzer --template code-analyzer

Allowed: read_file, code_analysis, query_db
Denied: write_file, change_permissions, forum_post

Data Bot

An agent for read-only analytics.

guardmesh init my-databot --template data-bot

Allowed: query_db, read_file
Denied: change_permissions, export_pii, write_file

Custom

A blank template for custom agents.

guardmesh init my-agent --template custom

Architecture

GuardMesh CLI integrates with:

• GuardMesh Registry: On-chain agent policy management
• GuardMesh Audit: Immutable decision log
• AXL Mesh: P2P guardian communication layer
• 0G Storage: Decentralized audit trail storage
• 0G Compute: TEE-verified guardian inference

┌─────────────┐
│   Your      │
│   Agent     │
└──────┬──────┘
       │ Intent
       ▼
┌─────────────────────────────────┐
│     Guardian Mesh (AXL)         │
│  ┌─────┐  ┌─────┐  ┌─────┐    │
│  │ G1  │  │ G2  │  │ G3  │    │ ◄── TEE Inference (0G Compute)
│  └─────┘  └─────┘  └─────┘    │
└──────┬──────────────────────────┘
       │ Consensus
       ▼
┌─────────────────────────────────┐
│   GuardMesh Contracts           │
│   • Registry (Policy)           │
│   • Audit (Decisions)           │
└──────┬──────────────────────────┘
       │ Merkle Root
       ▼
┌─────────────────────────────────┐
│   0G Storage                    │
│   • Intent Bundle               │
│   • Guardian Verdicts           │
│   • Verification Proof          │
└─────────────────────────────────┘

Configuration

GuardMesh CLI uses two configuration sources:

1. Local .env (project-specific)
2. Global config (~/.guardmesh/config.json)

Local configuration takes precedence.

Environment Variables

# Agent Configuration
AGENT_ID=my-agent-01
GUARDIAN_ENDPOINT=http://127.0.0.1:9002

# Blockchain Configuration
RPC_URL=https://evmrpc-testnet.0g.ai
REGISTRY_ADDRESS=0x...
AUDIT_ADDRESS=0x...

# Optional: Private key for signing
PRIVATE_KEY=0x...

Examples

Example 1: Code Review Agent

# Initialize
guardmesh init code-reviewer --template code-analyzer

# Register
guardmesh agent register \
  -i code-reviewer-01 \
  -r code_analysis_only \
  -a "read_file,code_analysis"

# Test
guardmesh test action \
  -a code-reviewer-01 \
  --action read_file \
  --target src/auth.ts

# Deploy
guardmesh deploy

Example 2: Analytics Bot

# Initialize
guardmesh init analytics-bot --template data-bot

# Register with strict permissions
guardmesh agent register \
  -i analytics-bot-01 \
  -r analytics_readonly \
  -a "query_db" \
  -d "change_permissions,export_pii"

# Test policy
guardmesh test policy -a analytics-bot-01

# Watch audit feed
guardmesh audit watch -a analytics-bot-01

Development

Build from source:

git clone https://github.com/yourusername/guardmesh-cli
cd guardmesh-cli
npm install
npm run build
npm link

Contributing

Contributions welcome! Please read our Contributing Guide.

License

MIT

Links

• GuardMesh Documentation
• 0G Network
• AXL Mesh
• Discord Community

Built with ❤️ for the 0G Hackathon - Best Agent Framework Track