no-patches

no-patches is a cursor rule for coding agents from LiuYihey/Agent-Engineering. It costs 279 tokens per session, scanned A, original, MIT.

A coding rule that says to fix the original cause of a bug instead of covering up its symptoms with one-off patches.

In plain words
What is it for?
Use it when designing or changing prompts, state handling, tool policies, and other code paths where repeated or incorrect behavior needs a lasting fix.
Why use it?
It reduces fragile fixes that work only for one screen, message, or failure case.

Cursor rule

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/liuyihey/agent-engineering/no-patches
Clone the repo
git clone --depth 1 https://github.com/LiuYihey/Agent-Engineering

Wrote 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.

agentmods badge for no-patches

README.md
[![agentmods](https://agentmods.dev/badge/rules/liuyihey/agent-engineering/no-patches.svg)](https://agentmods.dev/rules/liuyihey/agent-engineering/no-patches)
Your own site
<a href="https://agentmods.dev/rules/liuyihey/agent-engineering/no-patches"><img src="https://agentmods.dev/badge/rules/liuyihey/agent-engineering/no-patches.svg" alt="Measured on agentmods" height="20"></a>
Per session 279 This file is loaded in full into every session.
When invoked 279 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 $0.00279 $0.00279
Opus 5 $0.00139 $0.00139
Sonnet 5 $0.00056 $0.00056
Haiku 4.5 $0.00028 $0.00028

Measured 4d ago against content hash 729a252733ea, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

no-patches 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 4d ago.

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.

rules/no-patches.mdc · 25 lines

What it actually says

No Patches - Root-Cause Fixes Only

When something breaks or behaves wrong, fix the source (state model, prompt scope, persistence, tool policy). Do not paper over symptoms.

Avoid

  • Keyword/regex routing when the agent or UI already has conversation history and session state.
  • Dedupe, strip, trim, or merge logic whose job is to hide duplicate or wrong model output.
  • Forced tool calls or special-case branches that compensate for unclear prompts or missing context.
  • One-off examples or phrase lists embedded in prompts or code paths.

Prefer

  • Facts over inference rules: inject what the system knows; let the model or UI interpret the transcript.
  • Structural constraints: disable tools, phases, or code paths that should not run - not "please don't" in prose.
  • Single ownership: each artifact (plan, summary, segment) is produced once in the right layer; align server persistence with client rendering.
  • Prompt scope: if output is wrong or repeated, tighten role and phase boundaries (see concise-agent-prompts skill).

If a fix only works for one utterance, one screen, or one failure mode, redesign instead of extending the patch.

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. 4d ago First seen · 25 lines · 279 tokens per session scan A 729a252733ea

Subscribe to this mod's changes

no-patches is a cursor rule published in the GitHub repository LiuYihey/Agent-Engineering (5 stars, last pushed 1mo ago), licensed MIT. It adds 279 tokens to every session, about $0.0014 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.