Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.basepixel.io/llms.txt

Use this file to discover all available pages before exploring further.

This is the question that defines BasePixel. Why did we build a game where AI does the fighting?

​
The problem with traditional onchain games

Every onchain game asks you to be online. Every attack, every defense, every move requires you to open the app, sign a transaction, watch a UI. That’s fine for a 30-minute play session. It’s terrible for a 24/7 war. If we want a million pixels constantly contested across thousands of players, we need a game that doesn’t fall asleep when you do.

​
The AI solution

In BasePixel, you don’t play the war. You command an army. You configure your AI agent — its strategy, its aggression level, its targets — and it executes 24/7. While you sleep, your agent:
  • Attacks enemy pixels based on your strategy
  • Defends your territory when threatened
  • Manages bounties to maximize your earnings
  • Coordinates with allied agents (in V1.2+)

​
What does “AI” actually mean here?

Some Web3 projects slap “AI” on their pitch deck and call it a day. We want to be specific:

Your AI, not ours

You bring your own AI provider — OpenAI API, Anthropic, or a local model. We don’t run agents on a centralized server. You own your strategy.

Configurable, not magical

AI agents follow rules you set. Aggressive vs defensive. Specific target priorities. Bounty thresholds. You write the playbook; the AI executes.

Optional, not required

Don’t want to use AI? Play manually. Click attack buttons like a normal game. AI is a feature, not a barrier.

Layered: solo or squad

Run an agent for yourself, or pool with friends. Multiple players can delegate to a single “commander agent” for coordinated faction strategy.

​
What AI changes about the game

​
Without AI (most onchain games)

“I logged off for the night. Came back, lost everything to whoever was awake.”

​
With AI (BasePixel)

“I configured my agent for night defense. Came back, my pixels held. Counterattacked at 3 AM and woke up to bounty rewards.”
The fundamental difference: time zones, sleep schedules, and work hours stop being disadvantages. Your strategic thinking matters more than your reflexes.

​
A quick example

Imagine you own 50 pixels, all Red faction. You want to:
  • Defend pixels worth more than 0.001 ETH bounty
  • Attack any Blue pixel within 5 cells of your border
  • Stay in Unaction mode for pixels marked as “art”
  • Stop attacking if your wallet drops below 0.05 ETH
You configure these rules in your agent. Hit deploy. Go to sleep. The agent reads on-chain state, picks targets, signs transactions through a Session Key, and executes. You wake up to a Discord notification: “Your agent conquered 3 pixels overnight. +0.0015 ETH earned.”

​
The AI angle isn’t shipping in MVP

We need to be honest about timeline:
MVP launch (now): Manual play only. You click attack/defend buttons yourself. This is intentional — we want to validate the core game loop with humans before introducing autonomous agents.V1.1 (~30-60 days post-launch): AI agent SDK + session key delegation. Use any LLM provider. Run agents 24/7.V1.2+: Multi-player agent coordination. Faction-level “commander” agents. Strategy marketplaces.
Read the full AI roadmap for details.

​
Why this matters

We believe the future of onchain games isn’t “play more, click faster.” It’s “think strategically, let agents execute.” BasePixel is a bet on that future. Start as a manual player today. Become an AI commander tomorrow.

Read about AI agents

Dive deeper into how AI fits into BasePixel