Step 11 of 14
Nexus Intelligence
Nexus Intelligence is your research assistant for Bittensor mining decisions. Instead of manually checking dozens of subnets, reading Discord channels, and comparing metrics in spreadsheets — you ask Nexus a question and get a clear answer: MINE, SKIP, or WATCH. It does the heavy research so you can focus on making smart decisions about where to mine.
What is Nexus?
Nexus Intelligence is an autonomous research tool that lives on your VPS. It collects data from 5 different sources (explained below), crunches the numbers, and produces structured intelligence reports about any Bittensor subnet.
Think of each subnet as a small company you could invest your time and hardware in. Before investing, you'd want to know: How much does it pay? How many competitors are there? Is the community healthy? Is the code well-maintained? Nexus answers all of these questions automatically.
Important: Nexus runs through Agent Deck — you never launch it with raw terminal commands. It has its own group ("Nexus") and session ("chat") pre-configured for you.
The 5 Data Sources
Nexus doesn't guess — it pulls real data from 5 different sources and combines them into one report. Here's what each source provides and why it matters:
| Source | What it provides | Why it matters | Example |
|---|---|---|---|
| Metagraph | Live network data: miner count, validator count, emission rates, stake distribution | This is the real-time snapshot of a subnet. It tells you how many miners are competing right now, how much TAO is being distributed, and whether a few big miners are hogging all the rewards. | "Subnet 18 has 45 miners sharing 0.12 TAO/block" |
| Ralph (Taostats) | Historical performance data: trends, rank changes, emission history | Current data only tells half the story. Ralph shows you trends — is this subnet getting better or worse over time? Are miners earning more or less than last week? Trends help you predict the future. | "Emission per miner dropped 30% in the last 7 days" |
| Community Intelligence | Discord activity, developer engagement, community health signals | A healthy community means active developers who fix bugs, answer questions, and improve the subnet. A dead community means the subnet might be abandoned — risky to mine. | "Active Discord with 500+ members, devs responding within hours" |
| Code Analysis | Repository structure, code quality, update frequency, scoring mechanism complexity | The scoring mechanism is how the subnet decides which miners get rewarded. Clean, well-documented code means the rules are fair and predictable. Messy or complex code can hide traps that hurt new miners. | "Last commit 3 days ago, clean scoring function, well-documented" |
| Discord Monitoring | Real-time channel activity, announcement tracking, sentiment analysis | This catches breaking news: upcoming scoring changes, maintenance windows, rule updates. If the scoring algorithm is about to change, you want to know before you start mining, not after. | "Recent announcement: scoring update coming next week" |
Launching Nexus
Nexus is already set up in Agent Deck as its own group. Here's how to open it:
- Open Agent Deck — SSH into the VPS and type
agent-deck. This is your starting point for everything. - Navigate to the Nexus group — You'll see groups listed on the left side. Use the
↑↓arrow keys to highlightNexus. PressEnterto open it. - Select the chat session — On the right side, you'll see the
chatsession. Highlight it and pressEnter. - Nexus loads — You'll see a prompt where you can type questions in plain English. Just ask what you want to know about a subnet.
Ctrl+Q to return to Agent Deck, then select the other group. Nothing gets lost — both sessions keep running in the background.
Querying Subnets
Once inside Nexus, you simply type your question in plain English. You don't need special commands or syntax — just ask like you're talking to a colleague. Here are three common types of questions:
Quick check — when you want a fast overview of a subnet:
Deep analysis — when you want the full picture before making a decision:
Comparison — when you're deciding between two subnets:
After you ask, Nexus goes to work: it queries all 5 data sources, analyzes the results, and responds with a structured report. The report always includes a verdict (MINE, SKIP, or WATCH) and the data that supports it. You'll see specific numbers, trends, and explanations — not just a yes or no.
Understanding Verdicts
Every Nexus report ends with one of three verdicts. Think of them like traffic lights — green means go, red means stop, orange means wait and watch:
| Verdict | What it means | What you should do |
|---|---|---|
| MINE | This subnet is profitable and suitable for your setup. The competition is manageable, emissions are healthy, and the community is active. The data says this is a good opportunity right now. | Set up mining. Use Claude Code + GSD to clone the subnet repo, map the codebase, understand the scoring system, and deploy your miner. |
| SKIP | Not worth mining right now. Could be too competitive (too many miners splitting tiny rewards), low emissions (the subnet doesn't pay enough), or technical barriers (requires hardware you don't have). | Move on to other subnets. Don't waste time here. You can always come back later if conditions change. |
| WATCH | Interesting but not ready yet. The subnet shows potential — maybe emissions are rising, or the community just launched something new — but it's too early or too risky to jump in today. | Add it to your mental watchlist. Ask Nexus about it again in a week or two. If the trend continues improving, it might become a MINE. |
Key Metrics
Nexus reports include several important numbers. Here's what each metric actually means in plain terms, and how to tell if it's good or bad for you:
| Metric | What it actually means | Good sign | Bad sign |
|---|---|---|---|
| Emission per miner | How much TAO each miner earns on average per day. Think of it like the average salary at a company. Higher means more money for you. This is calculated by dividing the total subnet emissions by the number of active miners. | Above 0.01 TAO/day — enough to justify the electricity and hardware costs | Below 0.001 TAO/day — you'd spend more on electricity than you earn |
| Gini coefficient | This tells you how fairly the rewards are shared among miners. It's a number between 0 and 1. A low Gini (like 0.2) means everyone gets a fair share — rewards are spread evenly. A high Gini (like 0.9) means one or two top miners take almost everything while everyone else gets scraps. For a newcomer, low Gini is much better. | Below 0.5 — rewards are spread across many miners, giving newcomers a fair chance | Above 0.8 — a few dominant miners take most rewards. Very hard to compete as a newcomer |
| Community health | A qualitative score based on how active the subnet's Discord is, how often developers push updates, and whether the team responds to questions. A healthy community means the subnet is being maintained and improved — your mining setup won't suddenly break because of an unannounced change. | Active developers, regular code updates, helpful community members answering questions | Dead Discord with no messages for weeks, no code commits in months — the subnet might be abandoned |
| Competition type | What kind of hardware the subnet's scoring mechanism rewards. Some subnets need GPUs (graphics cards), others need CPUs, and some need specialized hardware. This tells you whether your RTX 4090 is the right tool for the job. | Matches your hardware — if the subnet rewards GPU work and you have an RTX 4090, you're set | Requires hardware you don't have — no point mining a CPU-optimized subnet with a GPU, or one that needs an A100 when you have a 4090 |
| Miner count trend | Whether the number of miners is going up, down, or staying stable over time. More miners = more competition = smaller slice of the pie for each person. This is like watching how many new restaurants open on the same street as yours. | Stable or slowly growing — competition is predictable, you can plan ahead | Rapid influx (sudden spike in new miners means a competition crunch is coming) or mass exodus (miners leaving usually means something is wrong — declining pay, upcoming changes, or technical issues) |
What-If Scenarios
One of Nexus's most powerful features is its ability to model hypothetical situations. Instead of guessing what would happen if you changed something, you can ask Nexus to simulate it. This lets you make decisions before spending money or time.
Budget planning — figure out if an investment makes sense before buying:
Nexus will estimate your potential earnings with the extra hardware and compare it to the cost. If the payback period is 6 months, that might be worth it. If it's 3 years, probably not.
Hardware comparison — compare options before committing:
Different hardware performs differently on different subnets. Nexus compares the expected performance so you can pick the best option for your budget.
Risk assessment — prepare for changes before they happen:
More miners means more competition and smaller rewards per person. Nexus calculates at what point your mining would stop being profitable, so you know your safety margin.
What-if scenarios help you make decisions before spending money. Always run a scenario before buying hardware or switching subnets — it takes 30 seconds to ask and could save you thousands.
Quick Reference
Keep this table handy. When you're not sure how to phrase a question for Nexus, just adapt one of these templates:
| What you want to know | What to ask Nexus |
|---|---|
| Is a subnet worth mining? | "Give me a report on subnet [N] with your recommendation" |
| Compare two subnets | "Compare subnet [A] and subnet [B] for mining with [your hardware]" |
| Check profitability | "What's the current emission per miner for subnet [N]?" |
| Plan a hardware purchase | "If I add [hardware], what's the expected return on subnet [N]?" |
| Monitor changes over time | "What changed on subnet [N] in the last 7 days?" |
Ready to go deeper? Next you'll learn the complete workflow for entering a subnet — from research to recommendation.