SLV — AI-Powered Solana Node Deployment

Deploy Solana validators, RPC nodes, and gRPC streaming infrastructure through natural conversation with AI agents.

SLV provides Ansible playbooks, Jinja2 templates, and AI agent skills that let you deploy and manage Solana nodes by simply describing what you need. No manual configuration files, no memorizing commands — just tell your AI agent what to do.

Quick Start — Deploy with AI

SLV CLI — The Fastest Way to Get Started

Install the CLI and launch the AI Console in 3 commands:

curl -fsSL https://storage.slv.dev/slv/install | sh
slv onboard    # Configure AI provider, API keys, and agent settings
slv c          # Launch AI Console — deploy nodes through conversation

The AI Console (slv c) gives you a rich terminal UI where specialist agents handle everything:

Cecil — Validator deployments (Jito, Agave, Firedancer)
Tina — RPC & gRPC Geyser deployments
Figaro — Server procurement and pricing
Setzer — Trading bots and Solana apps

Just describe what you need. The agent handles SSH setup, key generation, inventory creation, builds, snapshot download, and monitoring.

See slv.dev for full documentation.

ERPC Network Benefits

Servers purchased through erpc.global automatically get:

Dedicated snapshot endpoints — 7 global regions for fast node bootstrapping
Internal routing — dramatically lower bandwidth costs
Auto-detection — SLV automatically finds the nearest snapshot server via ping

For Developers

# Clone and run locally
git clone https://github.com/ValidatorsDAO/slv.git
deno task dev --help

# Build
deno task build

# Test
deno test -A

🚀 Just want to deploy? Run slv onboard then slv c — see SLV CLI below.

Prerequisites

ansible-core >= 2.15 (pip install ansible-core)
SSH access to target servers (key-based authentication)
solana-cli (optional, for local key generation)

Run the setup script to auto-install:

bash scripts/setup.sh

1. Install a Skill

SLV skills work with any AI coding agent — OpenClaw, Claude Code, Codex, Cursor, Windsurf, and more. Pick the method that fits your setup:

OpenClaw (via ClawHub)

npx clawhub install slv-validator

Claude Code

# Copy skill into your project
cp -r dist/oss-skills/slv-validator /your/project/.claude/skills/

Codex / Cursor / Other AI Agents

# Add SKILL.md to your agent's context
cp dist/oss-skills/slv-validator/SKILL.md /your/project/AGENTS.md
# Or reference it in your agent's config

Direct from GitHub

# Clone just the skill you need
git clone --depth 1 https://github.com/ValidatorsDAO/slv.git
cp -r slv/dist/oss-skills/slv-validator ./my-skill

Skills are just Markdown files (SKILL.md + AGENT.md) with Ansible playbooks — any AI that can read files and run commands can use them.

2. Talk to Your Agent

You: Deploy a mainnet Jito validator on 203.0.113.10

Agent: I'll set up a mainnet Jito validator. Let me walk you through the configuration:
       - Server IP: 203.0.113.10
       - What SSH user should I use? (default: solv)
       - Do you have existing validator identity and vote account keys,
         or should we generate new ones?
       ...

The AI agent guides you through the entire process interactively:

Collects configuration — server details, validator type, keys, network settings
Validates inputs — checks SSH connectivity, key formats, version compatibility
Generates inventory — creates Ansible inventory from your responses
Offers dry-run — shows what will happen before executing
Deploys — runs the appropriate Ansible playbooks
Monitors — tracks startup progress and slot sync

3. Manage Your Nodes

You: Restart the validator on 203.0.113.10
You: Update Solana to v3.1.8 on my RPC node
You: What's the slot sync status?
You: Switch validator identity with zero downtime

Available Skills

Skill	ClawHub	What It Does
slv-validator	`npx clawhub install slv-validator`	Deploy & manage mainnet/testnet validators (Jito, Agave, Firedancer)
slv-rpc	`npx clawhub install slv-rpc`	Deploy & manage RPC nodes (Standard, Index, Geyser gRPC, Index+gRPC)
slv-grpc-geyser	`npx clawhub install slv-grpc-geyser`	Deploy & manage gRPC Geyser streaming (Yellowstone, Richat)

Each skill includes:

SKILL.md — Complete playbook knowledge for the AI agent (works with any AI)
AGENT.md — Interactive deployment flows and behavior rules
scripts/setup.sh — Auto-install prerequisites (ansible-core, SSH, solana-cli)
ansible/ — Ansible playbooks (can also be used standalone)
examples/ — Sample inventory files

No lock-in. Skills are plain Markdown + Ansible. Use them with any AI agent, or run the playbooks directly without AI.

How It Works

SLV uses Ansible playbooks and Jinja2 templates to deploy Solana nodes. The AI agent skills wrap this infrastructure with conversational interfaces:

User request → AI Agent (SKILL.md knowledge) → Ansible playbooks → Target server

Key design principles:

Remote-only — all configuration from your local machine, no direct node logins
Dummy key start — validators always start with an unstaked identity, then switch
Source builds — Solana binaries built from GitHub source (no pre-built downloads)
Firewall-first — SSH IP restrictions and nftables configured during init

Using Without AI (Direct Ansible)

You can also use the playbooks directly:

cd dist/oss-skills/slv-validator/ansible/

# Deploy a mainnet Jito validator
ansible-playbook -i inventory.yml mainnet-validator/init.yml \
  -e '{"validator_type":"jito","solana_version":"v3.1.8-jito","snapshot_url":"https://..."}'

# Restart a validator
ansible-playbook -i inventory.yml mainnet-validator/restart_node.yml

# Build Solana from source
ansible-playbook -i inventory.yml cmn/build_solana.yml \
  -e '{"solana_version":"v3.1.8"}'

Community

Contributing

Bug reports and pull requests are welcome. This project follows the Contributor Covenant code of conduct.

License

Apache-2.0