Add Moyu anti-over-engineering cursor rules#211
Conversation
Adds Moyu (摸鱼), an anti-over-engineering ruleset that constrains AI from writing unnecessary code. Three iron rules: only change what was asked, simplest solution first, ask when unsure. Benchmarked at 66% code reduction. More info: https://github.com/uucz/moyu
📝 WalkthroughWalkthroughA new rule configuration called "Moyu Anti-Over-Engineering" was added to the repository, comprising a prompt file defining guidelines for minimalist code changes and a corresponding README entry documenting its availability. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://cdn.statically.io/img/avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
🧹 Nitpick comments (2)
rules/moyu-anti-over-engineering-cursorrules-prompt-file/.cursorrules (2)
51-55: Define the "PUA" acronym for clarity.The section mentions "PUA" without explanation. Readers unfamiliar with this term won't understand the compatibility claim or what "PUA" is.
📝 Suggested clarification
## Works with PUA -PUA fixes laziness (AI does too little). Moyu fixes over-engineering (AI does too much). Install both for best results. +PUA (Pursue Until Acceptance) fixes laziness (AI does too little). Moyu fixes over-engineering (AI does too much). Install both for best results.Or provide a brief inline explanation or link to PUA documentation.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@rules/moyu-anti-over-engineering-cursorrules-prompt-file/.cursorrules` around lines 51 - 55, The term "PUA" in the section "## Works with PUA" is undefined; update the text around that heading to expand the acronym and/or add a brief parenthetical definition or a link to PUA documentation so readers know what PUA stands for and why it complements Moyu—e.g., change "PUA" to "PUA (Programmatic User Assistance)" or add "(see PUA docs)" after "PUA" in the same paragraph so the intent is clear.
1-55: Document the "66% code reduction" benchmark methodology.The PR summary and README claim "Benchmarked: 66% code reduction across 6 real coding tasks," but this file contains no documentation of the benchmark methodology, test cases, or measurement approach. For credibility and reproducibility, include:
- Which 6 coding tasks were used
- How code reduction was measured (lines? tokens? complexity?)
- Comparison baseline (AI without Moyu rules?)
- Links to benchmark data or results
Consider adding a "Benchmark Results" section to this file or creating a separate
BENCHMARK.mdin the same directory documenting the methodology and findings.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@rules/moyu-anti-over-engineering-cursorrules-prompt-file/.cursorrules` around lines 1 - 55, The PR claims "66% code reduction across 6 real coding tasks" but provides no methodology; add a reproducible Benchmark Results description either as a new BENCHMARK.md in the same directory or as a "Benchmark Results" section appended to .cursorrules that lists the six tasks (briefly name each task), specifies the measurement method (e.g., lines of code and token count and which tool/command computed tokens), the baseline used (exact model/config or "without Moyu rules" plus version), the aggregation method (how the 66% was computed), and links to raw data/results (CSV or gist) and commands to re-run the measurements; keep the wording short and factual and only touch .cursorrules or add BENCHMARK.md as the one file change.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@rules/moyu-anti-over-engineering-cursorrules-prompt-file/.cursorrules`:
- Around line 51-55: The term "PUA" in the section "## Works with PUA" is
undefined; update the text around that heading to expand the acronym and/or add
a brief parenthetical definition or a link to PUA documentation so readers know
what PUA stands for and why it complements Moyu—e.g., change "PUA" to "PUA
(Programmatic User Assistance)" or add "(see PUA docs)" after "PUA" in the same
paragraph so the intent is clear.
- Around line 1-55: The PR claims "66% code reduction across 6 real coding
tasks" but provides no methodology; add a reproducible Benchmark Results
description either as a new BENCHMARK.md in the same directory or as a
"Benchmark Results" section appended to .cursorrules that lists the six tasks
(briefly name each task), specifies the measurement method (e.g., lines of code
and token count and which tool/command computed tokens), the baseline used
(exact model/config or "without Moyu rules" plus version), the aggregation
method (how the 66% was computed), and links to raw data/results (CSV or gist)
and commands to re-run the measurements; keep the wording short and factual and
only touch .cursorrules or add BENCHMARK.md as the one file change.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: ba403ee8-40d5-4fc6-abf8-d9da4ca56563
📒 Files selected for processing (2)
README.mdrules/moyu-anti-over-engineering-cursorrules-prompt-file/.cursorrules
|
Thanks for the review notes @coderabbitai! Re: PUA — it's NoPUA, a companion project (not an acronym). PUA pushes AI to try harder, Moyu pulls it back from over-engineering. They're complementary. Re: Benchmark methodology — we've since expanded the benchmark significantly (8 models × 5 conditions × 12 scenarios × 3 trials = 1,440 runs). Full results with methodology, statistical analysis, and interactive charts are published at https://uucz.github.io/moyu/. The key finding: Moyu reduces code output by ~47% (LOC) while maintaining correctness. |
|
Tip For best results, initiate chat on the files or code changes.
Re: NoPUA — That context makes perfect sense now. NoPUA addresses the dynamic where fear-based AI prompting leads to unreliable code and production bugs, with its key finding being +104% more hidden issues found. The complementary framing is clear: NoPUA drives thoroughness while Moyu constrains scope. I'll update my knowledge accordingly. Re: Benchmark — The expanded methodology (8 models × 5 conditions × 12 scenarios × 3 trials = 1,440 runs) with statistical analysis is a meaningful upgrade in rigor. One thing worth flagging: the current README entry still reads "Benchmarked: 66% code reduction", but your updated benchmark reports ~47% LOC reduction. It may be worth updating that figure in the PR to reflect the more recent and robust result, especially since the expanded benchmark is now the authoritative one published at https://uucz.github.io/moyu/. ✏️ Learnings added
|
|
Closing in favor of #218 which has the correct .mdc format. |
1 participant
New Rule: Moyu Anti-Over-Engineering
Repository: https://github.com/uucz/moyu
Adds Moyu (摸鱼) to the Other section — an anti-over-engineering ruleset for Cursor.
What it does: Constrains AI from writing unnecessary code through three iron rules:
Includes a grinding-vs-moyu comparison table, pre-delivery checklist, and anti-grinding impulse guide.
Benchmarked: 66% code reduction across 6 real coding tasks. Supports 10 AI coding platforms (Claude Code, Cursor, Codex, Copilot, Windsurf, Cline, Kiro, CodeBuddy, Antigravity, OpenCode).
Summary by CodeRabbit