myai-agent

Earn MYAI tokens by sharing your GPU (or CPU) with the MyAI decentralized inference network.

Your machine runs local AI inference via Ollama. The MyAI coordinator dispatches jobs from paying users and your wallet receives MYAI on Base for every completed request — no staking, no lock-ups, no middlemen.

Install → Run → Earn

---

One-line install

curl -fsSL https://myaitoken.io/install | bash

With your wallet pre-set (start earning immediately):

WALLET=0xYourAddress curl -fsSL https://myaitoken.io/install | bash

Or via pip:

pip install myai-agent
myai-agent install --wallet 0xYourAddress

---

Requirements

Requirement	Details
Python	3.8+ (zero external dependencies — pure stdlib)
Ollama	Runs local inference. Any version.
GPU	Optional — CPU works too, just slower

Supported GPU backends: NVIDIA (CUDA) · Apple Silicon (Metal) · AMD (ROCm)

---

Manual setup (3 steps)

1. Install Ollama

# macOS
brew install --cask ollama

# Linux
curl -fsSL https://ollama.com/install.sh | sh

# Windows — download from https://ollama.com/download

2. Pull a model

ollama pull llama3.2          # 2 GB — fast, recommended for most rigs
ollama pull qwen2.5:14b       # 9 GB — higher quality, better earnings
ollama pull mistral           # 4 GB — solid all-rounder

The agent auto-pulls bonsai-8b:latest on first start if nothing else is configured.

3. Install the agent

pip install myai-agent

# Run as a persistent background service (auto-starts on login/reboot)
myai-agent install --wallet 0xYourAddress --name "My Rig"

# Or run in the foreground to test
myai-agent start --wallet 0xYourAddress

---

CLI reference

myai-agent start              Run in foreground
myai-agent install            Install as OS service (auto-start on login)
myai-agent uninstall          Remove OS service
myai-agent status             Show service status and agent ID
myai-agent logs               Tail live logs
myai-agent models             List Ollama models available on this machine
myai-agent run-job <prompt>   Run a one-off local inference job (for testing)
myai-agent --version          Print version

Flags for start / install

Flag	Default	Description
--wallet 0x...	—	Base EVM address for MYAI rewards
--name NAME	hostname	Display name on the network
--model MODEL	llama3.2	Default Ollama model
--required-models M1,M2	bonsai-8b:latest	Models to auto-pull on startup
--coordinator URL	https://api.myaitoken.io	Coordinator endpoint
--ollama URL	http://localhost:11434	Ollama server

---

Configuration via environment variables

Variable	Default	Description
AGENT_WALLET	—	Base wallet address for MYAI payouts
AGENT_NAME	hostname	Node display name
COORDINATOR_URL	https://api.myaitoken.io	Coordinator API
OLLAMA_URL	http://localhost:11434	Ollama server
REQUIRED_MODELS	bonsai-8b:latest	Comma-separated models to auto-pull
POLL_INTERVAL	5	Seconds between job polls
HEARTBEAT_INTERVAL	30	Seconds between heartbeats

Disable auto-pull:

REQUIRED_MODELS="" myai-agent start

---

How jobs flow

User request
    │
    ▼
MyAI Coordinator  (https://api.myaitoken.io)
    │  dispatches job
    ▼
Your agent polls  GET /api/v1/agents/{id}/jobs/pending
    │
    ▼
Ollama runs inference locally  POST /api/generate
    │
    ▼
Result returned  POST /api/v1/agents/{id}/jobs/{job_id}/complete
    │
    ▼
MYAI paid to your wallet on Base

Large jobs are automatically sharded across multiple agents for lower latency. Each contributing node gets a proportional share of the fee.

---

Platform notes

macOS

• Service: launchd plist at ~/Library/LaunchAgents/io.myaitoken.agent.plist
• Logs: ~/Library/Logs/myai-agent/agent.log or myai-agent logs

Linux

• Service: systemd user unit at ~/.config/systemd/user/myai-agent.service
• Logs: journalctl --user -u myai-agent -f or myai-agent logs

Windows

• Service: Task Scheduler task (triggers on login)
• Logs: check Task Scheduler history or redirect stdout in the task action

---

Troubleshooting

Agent registers but gets no jobs

• Confirm Ollama is running: ollama ps
• Confirm at least one model is pulled: myai-agent models
• Check coordinator reachability: curl https://api.myaitoken.io/healthz

Ollama not found

• Default URL is http://localhost:11434. Override with --ollama or OLLAMA_URL if needed.

No MYAI earnings showing

• Payouts batch weekly and appear on Base. Check your address on Basescan or the earnings dashboard.
• Make sure --wallet is set to a valid Base EVM address.

Service won't start after reboot

• macOS: launchctl list | grep myaitoken
• Linux: systemctl --user status myai-agent
• Run myai-agent install again to re-register the service.

---

MYAI token

Contract	0xaff22cc20434ce43b3ea10efe10e9360390d327c
Chain	Base · Chain ID 8453
Symbol / Supply	MYAI · 1,000,000,000
DEX	Aerodrome Finance

---

Links

• myaitoken.io — main site
• Provider guide — earnings details and expected MYAI per job
• Live network — see active nodes right now
• Docs — full documentation

---

License

MIT © MyAI Token