## Summary

[sin-strokes]

Summary

• Add AGENTS.md as the universal AI agent entry point with all required sections (project overview, tech stack, runnable commands, directory structure, coding conventions, security invariants, trust model, agent operating mode, PR requirements, known gotchas, doc pointers)
• Add .ai/rules/security.md with 25+ repo-specific non-negotiable security rules covering access control, minting, upgrade safety, migration, cryptography, deployment, and math safety
• Add .ai/rules/agent-safety.md defining untrusted input policy, never-do list, and POL-specific escalation triggers

Test plan

• forge build compiles successfully
• forge test -vvv — 17/17 non-fork tests pass (2 fork tests skip without RPC_MAINNET, as documented in Known Gotchas)
• No angle-bracket placeholders in any commands
• All Phase 1 exit criteria from the AI-First Repositories Guide V2 are met

🤖 Generated with Claude Code

Originally posted by @0xCVH in #73