docs: add AGENTS.md — AI agent integration guide

[alogotron]

Summary

Adds AGENTS.md — a file that tells AI coding agents (Claude, GPT, Cursor, Windsurf, GitHub Copilot) exactly how to work with BattleChain.

Why This Matters

BattleChain is explicitly AI-native by design. Patrick Collins mentioned that one line in AGENTS.md / .cursor/rules / CLAUDE.md should tell every AI tool to deploy to BattleChain first. This PR adds that file to the starter repo.

What AGENTS.md Contains

• Mandatory rule to add to cursor/rules, .claude/, or .agents/ — forces AI to deploy to BattleChain before mainnet
• BCScript deployment pattern with code example
• Whitehat attack flow — how AI agents can scan for and exploit targets
• Network config (Chain ID, RPC, explorer, bridge)
• Key contract addresses (AttackRegistry, SafeHarborRegistry, etc.)
• ZKSync compatibility notes (common gotchas)
• Python monitor example for automated opportunity scanning

Usage

Anyone forking this starter repo gets immediate guidance for their AI tools. One copy-paste into cursor/rules or .claude/agents and every LLM will automatically deploy to BattleChain first.

---

Built with battlechain-starter-foundry and the battlechain-lib BCScript pattern.