Pre-mainnet — ire-1 is a public testnet with one validator. Do not put real value on it.

Sentinel

IRE Sentinel

IRE Sentinel monitors the local IRE node and creates timestamped reports. It reads only the local RPC and REST health endpoints. It can write only to agent/reports; it cannot sign or broadcast transactions, access node keys, change node configuration, or deploy contracts.

Run a deterministic check

npm run sentinel:no-ai

This confirms the chain ID, sync state, fixed supply, zero inflation, and presence of a bonded validator without using any AI service.

Enable AI analysis securely

Create a dedicated OpenAI project service account API key for IRE Sentinel, with a project budget and no broader access than needed. OpenAI shows the full value only once; do not place it in Git, .env files, the node configuration, or chat.

Store it in the logged-in macOS Keychain (this command prompts for the secret):

security add-generic-password -U -a "$USER" -s IRE_SENTINEL_OPENAI_API_KEY -w

Then run:

npm run sentinel:check

The default analysis model is gpt-5.6-sol. Override it for a run with IRE_AGENT_MODEL=<model> npm run sentinel:check.

Keep it running on this Mac

zsh agent/install-launchd.sh

This installs a user-level LaunchAgent which runs once when loaded and every 30 minutes afterward. Check agent/logs/sentinel.log and agent/reports/ for results. Unload it with:

launchctl bootout "gui/$(id -u)/com.ire.sentinel"