Rules, prompts, and a security checklist that make Cursor, Claude Code, and other coding agents build trackers, alert bots, and agents for crypto users without the classic disasters: a private key in the repo, a bot token in a log, an RPC ban, or a “safe” label on data that was never fetched.

What’s inside
• 6 Cursor rules (.cursor/rules): core, secrets and keys, Solana RPC and API limits, Telegram/Discord bots, trackers and data, testing and shipping
• CLAUDE.md and AGENTS.md with the same rules for Claude Code, Codex, Windsurf, Aider, and OpenClaw-style agents
• 12 starter prompts: scaffold a read-only tracker, add alerts, harden the RPC client, security review, debug a 429 storm, and more
• 8-section security checklist: keys, secrets, rate limits, untrusted input, bots, data, dependencies, shipping
• scan_secrets.py: finds Solana keypairs, seed phrases, bot tokens, webhooks, API keys, and keyed RPC URLs, plus a git pre-commit hook
• Templates: .env.example, .gitignore, and a rate-limited JSON client

Setup: copy the folders into your repo. Takes about 2 minutes. No installs.

Version 1.0 · Last updated Nov 12, 2026.
Reduces common risks. Not a security audit, not financial advice.