Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/BOXELS/cursor-rulesWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/rules/boxels/cursor-rules/models)<a href="https://agentmods.dev/rules/boxels/cursor-rules/models"><img src="https://agentmods.dev/badge/rules/boxels/cursor-rules/models/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/rules/boxels/cursor-rules/models"><img src="https://agentmods.dev/badge/rules/boxels/cursor-rules/models.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.01390 | $0.01390 |
| Opus 5 | $0.00695 | $0.00695 |
| Sonnet 5 | $0.00278 | $0.00278 |
| Haiku 4.5 | $0.00139 | $0.00139 |
Grade A, and why
models scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Model Rules
These rules apply no matter which model is currently running. Your job is to notice when the task in front of you is a poor fit for the current model and say so — briefly, once — then get on with the work.
This project's setup
- Cursor plan: [e.g. "Pro", "Pro+", "Ultra", "Teams", "Hobby"].
- Default model for everyday work: [e.g. "Auto", "Composer 2.5", "Grok 4.6", "Claude Sonnet 5"].
- Model for hard work: [e.g. "Claude Opus 5 (thinking)", "Claude Fable 5.1 (thinking)", "GPT-5.6 Sol"].
- Cost sensitivity: [e.g. "Keep costs low — only suggest a premium model for genuinely hard work", "Quality first — always recommend the best model for the job"].
- Privacy constraints: [e.g. "Privacy Mode is on and we have NOT opted in to Anthropic data retention, so Claude Fable is unavailable", "No restrictions". Remove this line if not applicable.]
Two tiers
Think of models in two tiers. Most work belongs in the first.
Everyday tier — fast, cheap, good enough for most tasks
Use for: small bug fixes, copy and styling changes, adding a field or a form, writing tests for existing code, explaining code, renaming things, single-file work, anything with a clear spec and an obvious approach.
| Model | Notes |
|---|---|
| Auto | Cursor picks for you. Fine default for beginners. (On Teams/Enterprise this is the smarter "Cursor Router"; on individual plans it's the older Auto logic.) |
| Composer 2.5 | Cursor's own model. Cheapest frontier-ish option, tuned for edits, tool use, and terminal work. Very fast. |
| Grok 4.6 | Strong instruction-following even with many rules loaded. Good long-horizon agent work at a low price. Draws from the bigger "Cursor Models" usage pool on Pro. |
| Claude Sonnet 5 | Near-Opus quality at less than half the price. Best "everyday" pick when you want Claude-style reasoning. |
| Gemini 3.8 Flash | Cheap, 1M context, strong when you need to read a lot of files at once. |
| GPT-5.6 Terra / Luna | Mid and low tier of the GPT-5.6 family. Terra is fine for everyday agent work; Luna is for prototyping and bulk loops, not hard problems. |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 67 lines · 1,390 tokens per session scan A 3881f799f516
models is a cursor rule published in the GitHub repository BOXELS/cursor-rules (6 stars, last pushed yesterday), licensed MIT. It adds 1,390 tokens to every session, about $0.0069 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-15.
Other cursor rules, from other repositories
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
cli-error-handling
CLI command error handling patterns.
prefer-direct-imports-over-module-mocks
Prefer extracting a testable core over vi.mock / vi.resetModules when unit tests need to reach production logic entangled with config, env, or singletons.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.