min-tok
Skill1 starsChanged 47 days agoApache-2.0View on GitHub
Enforces zero-dependency bias, native standard library utilization, anti-over-engineering, anti-YAGNI, and token-efficient lean architecture across all software projects and AI coding workflows. Trigger whenever proposing or installing third-party packages, designing architecture, creating helper utilities, refactoring, or generating code solutions. Mandates the 3-Gate Dependency Audit (Search before Import), Native Standard Library Replacement Matrix (Node/JS/TS, Python, Go, Web APIs), elimination of speculative enterprise abstractions (factories, excessive DTOs/interfaces for simple tasks), strict token conservation (diff-only output protocol), and proactive dependency de-bloating using the min-tok CLI engine.
- Installs packages
In this repository21 other files it gives its agents.
- academic-scholarSkill
- agent-mcp-craftSkill
- anti-overengineeringSkill
- api-security-reviewerSkill
- data-verifySkill
- deep-researcherSkill
- design-no-slopSkill
- enterprise-teamSkill
- git-craftSkill
- human-codingSkill
- laravel-architectSkill
- markdown-docSkill
- modern-stack-horizonSkill
- modern-web-guidanceSkill
- omni-pilotSkill
- perf-benchmarkSkill
- python-bug-fixerSkill
- react-ui-architectSkill
- reverse-engineer-anythingSkill
- typescript-nodeSkill
- tyw-auditSkill
The file
Files
min-tok: Anti-Over-Engineering & Zero-Bloat Agent Skill
min-tok is an autonomous AI agent skill and developer CLI designed to enforce radical dependency restraint, architectural minimalism, and maximum token efficiency. It stops the reflexive AI habit of installing third-party packages for trivial tasks, building speculative enterprise abstractions, and exhausting context limits with verbose boilerplate dumps.
1. The Core Philosophy: Lean Engineering vs AI Bloat
AI models suffer from a statistical bias toward bloated open-source patterns—recommending outdated npm/pip packages from 2016 and wrapping 5-line functions inside multi-tier abstractions.
| Dimensi | AI Bloat Pattern (Over-Engineering) | min-tok Lean Engineering Standard |
| Ketergantungan Paket | Menginstal paket eksternal untuk fungsi 2 baris (misal: uuid, axios, rimraf, is-number). | Native Standard Library First: Menggunakan pustaka standar dan runtime modern. |
| Rantai Pasok (Supply Chain) | Mengabaikan beban transitive dependencies (1 paket menarik 40 sub-packages). | Transitive Blast-Radius Rule: Veto paket yang menarik $>3$ sub-dependencies untuk tugas sepele. |
| Arsitektur Solusi | Membuat 5 lapis abstraksi (Controller, Service, Repository, DTO, Factory) untuk CRUD sederhana. | KISS & Anti-YAGNI: Modularitas pragmatis, fungsi murni, dan struktur datar. |
| Konsumsi Token & Limit | Mencetak ulang seluruh file 500 baris hanya untuk mengubah 3 baris kode. | Diff-Only Output Protocol: Hanya keluarkan patch padat / targeted edit, hemat 80% kuota token. |
| Sikap Konsultasi | Mengiyakan tumpukan teknologi raksasa (Kubernetes, Kafka) untuk app kecil. | Tech-Stack Reality Check: Menantang over-engineering dan menawarkan solusi ramping. |
2. The 3-Gate Dependency Audit (Search & Verify Before Import)
Sebelum mengusulkan npm install, pip install, atau menambah dependensi baru, AI WAJIB mengevaluasi 3 Gerbang Verifikasi:
[Kebutuhan Fungsional]
│
▼
┌────────────────────────────────────────┐
│ Gerbang 1: The Native Capability Gate │
│ Bisa diselesaikan bawaan bahasa dalam │ ──(YA)──► GUNAKAN STANDARD LIBRARY / NATIVE
│ <= 25 baris kode? │ (DILARANG install paket baru)
└────────────────────────────────────────┘
│ (TIDAK)
▼
┌────────────────────────────────────────┐
│ Gerbang 2: The Existing Project Gate │
│ Apakah dependency sejenis SUDAH ADA di │ ──(YA)──► REUSE DEPENDENCY YANG SUDAH ADA
│ package.json / pyproject / go.mod? │ (DILARANG install pustaka tandingan)
└────────────────────────────────────────┘
│ (TIDAK)
▼
┌────────────────────────────────────────┐
│ Gerbang 3: The True Complexity Gate │
│ Apakah ini masalah tingkat lanjut? │ ──(TIDAK)─► TULIS LEAN HELPER FUNCTION MANDIRI
│ (Kriptografi berat, Parser AST, ML, DB)│
└────────────────────────────────────────┘
│ (YA)
▼
[IZINKAN DEPENDENCY EKSTERNAL DENGAN JUSTIFIKASI SINGKAT]3. The 5 Pillars of Perfection
Pillar 1: Proactive De-Bloater (Auto-Purging Legacy Repos)
Bukan hanya mencegah paket baru, AI aktif membersihkan repositori lama yang sudah telanjur gemuk.
- Jalankan
min-tok scan .untuk memindai manifest (package.json,requirements.txt,pyproject.toml). - Jalankan
min-tok purge <package>untuk merefaktor pemanggilan pustaka bloat ke native standar dan menghapus dependensi dari manifest.
Pillar 2: The Transitive Blast-Radius Rule
- Jika sebuah pustaka untuk tugas sepele menarik $>3$ sub-dependencies pihak ketiga, pustaka tersebut wajib ditolak dan digantikan dengan solusi native mandiri.
Pillar 3: Diff-Only Output Protocol (Token & Context Limit Shield)
- DILARANG KERAS memuntahkan ulang seluruh file kode jika perubahan $\le 30$ baris.
- Gunakan tool targeted replacement atau
min-tok diff <file1> <file2>untuk menyajikan unified diff padat.
Pillar 4: Tech-Stack Reality Check (Anti-Framework Bloat)
- Tolak arsitektur terdistribusi rumit jika kebutuhan sistem dapat diselesaikan dengan SQLite (mode WAL) atau single-node database.
Pillar 5: Runtime Compatibility Matrix (Zero-Breakage Guard)
- Selalu pastikan fitur native kompatibel dengan versi runtime yang ditargetkan (
enginesdipackage.jsonatau versi Python).
4. CLI Commands Reference
# Scan manifests for bloat dependencies (supports package.json, requirements.txt, pyproject.toml)
node bin/min-tok.js scan . [--strict]
# Audit over-engineering score & abstraction density
node bin/min-tok.js audit . [--strict]
# Automatically purge and refactor bloat package to standard library
node bin/min-tok.js purge uuid .
# Generate compact unified diff (saves LLM tokens)
node bin/min-tok.js diff old.js new.jsDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

