mint-cookbook
MindLab-Research/mint-cookbook
This repository provides comprehensive instructions, architectural guidelines, and scaffolds for managing AI research experiments and model reproduction workflows within a monorepo.
7 3No licenseUpdated 2026-07-21
69
GoodOverview
The mint-cookbook skill defines a rigorous framework and set of operational rules for AI researchers and engineers working inside the MindLab-Research/mint-cookbook monorepo. It establishes strict layer separation for agent context, standardization for experiment scaffolds (like training scripts, data splits, and evaluation entrypoints), and best practices for maintaining reproducibility, artifact management, and metric reporting across various ML training methods such as SFT, DPO, and GRPO.
Capabilities
- ▸Defines a structured monorepo workflow for machine learning experiments
- ▸Provides rules for single-file experiment runtimes (train.py) with dry-run and eval-only support
- ▸Establishes dataset split conventions and artifact management for train and eval pipelines
- ▸Enforces strict version control, documentation synchronization, and registry updates
- ▸Standardizes scaffold profiles for SFT, DPO, GRPO, and evaluation harnesses
Best for
Setting up and scaffolding new AI research experiments or paper reproductions, Enforcing consistent architectural patterns and CLI flag conventions across ML training runs, Managing structured dataset pipelines, smoke testing tiers, and evaluation paths using uv and python scripts, Coordinating multi-layer context and task packages for AI coding agents within a research monorepo
Works with
Claude CodeCursorGitHub CopilotCodexChatGPTGemini CLIOpenCode
Evoa Score breakdown= Σ (score × weight)
69Task usefulness20%88 → +17.6
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%90 → +13.5
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%100 → +15.0
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%37 → +4.4
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%100 → +15.0
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%85 → +8.5
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%29 → +2.3
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%100 → +5.0
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total81.4 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
The skill enforces credential-free validation via dry-run mechanisms and restricts local machine paths or personal routing from being checked into open-source contract files.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.