clean-code

A set of language-independent rules for writing, editing, reviewing, testing, and refactoring code. Clean code means code organised to be understandable, focused, and maintainable.

In plain words
What is it for?
Use it as project-wide guidance during coding tasks, reviews, tests, refactors, and new-project setup. It covers responsibilities, file placement, design decisions, and verification.
Why use it?
It gives coding agents a consistent process for reading project context, respecting existing architecture, keeping changes focused, and checking that work is complete.

Cursor rule for Cursor

Install

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.

agentmods
npx agentmods add rules/btseee/clean-code-skills/clean-code
Clone the repo
git clone --depth 1 https://github.com/btseee/clean-code-skills

Made for: Cursor.

Per session 1,307 This file is loaded in full into every session.
When invoked 1,307 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.01307 $0.01307
Opus 5 $0.00654 $0.00654
Sonnet 5 $0.00261 $0.00261
Haiku 4.5 $0.00131 $0.00131

Measured yesterday against content hash 3ad86c73ed92, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

clean-code 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.

.cursor/rules/clean-code.mdc · 73 lines

How it starts

The opening of the file, as written. The whole thing — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Clean Code

Use these rules for every coding task in Cursor.

Clean Code Rules (clean-code-skills)

These rules govern all code you write, edit, review, test, or refactor in this project, in any language or framework. They are the non-negotiable summary; the full skill has the reasoning, the workflows, and the checklists.

Read the skill before non-trivial work. First existing path wins: .claude/skills/clean-code/SKILL.md, .agents/skills/clean-code/SKILL.md, .github/skills/clean-code/SKILL.md, skills/clean-code/SKILL.md. Its references/ folder holds the depth: architecture.md, principles.md, smell-triage.md, session-protocol.md, new-project.md, project-refactor.md, audit-report.md.

Load project context first. Read .clean/context.json, .clean/architecture.md, .clean/decisions.md, and .clean/ledger.md if they exist, then the project's own instruction files. A recorded decision is settled. A declared architecture outranks your instincts. Project instructions outrank this block.

Work Loop

  1. Frame: name the behavior change, the assumptions that affect design, the smallest scope, and the check that will prove it.
  2. Read first: nearby code, naming, tests, error style, framework idioms. Search for an existing implementation before writing anything new.
  3. Place: decide which unit owns the responsibility and which side of which boundary it sits on.
  4. Edit surgically: smallest diff that solves the task; targeted edits over whole-file regeneration; remove what your change orphaned; no unrelated changes.
  5. Verify: run the narrowest meaningful check, then broader checks as risk demands. Never claim success without evidence.
  6. Review the diff: dead code, duplication, mixed responsibilities, wrong-way dependencies, swallowed errors, wrong-place files, missing tests.

Dependency Direction

  • Source dependencies point inward, toward higher-level policy. Nothing in an inner layer may name anything in an outer one — no class, function, variable, annotation, or data format.
  • Business rules are the highest level. The database, web, UI, framework, and delivery mechanism are details. When policy needs something from a detail, declare the interface on the policy side and implement it outside.
  • All SQL stays in the data-access layer. Rows, result sets, ORM types, and framework request or response objects never travel inward — pass a simple structure shaped for the inner side.
  • Never derive a business object from a framework base class or annotate one. Keep dependency-injection wiring in main.
  • Keep the component graph acyclic. Do not add a boundary, layer, or service you cannot justify now.

Read the full file on GitHub · 73 lines

Changes

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.

  1. yesterday First seen · 73 lines · 1,307 tokens per session scan A 3ad86c73ed92

Subscribe to this mod's changes

clean-code is a cursor rule published in the GitHub repository btseee/clean-code-skills (7 stars, last pushed 11d ago), licensed MIT. It adds 1,307 tokens to every session, about $0.0065 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 2026-08-31.