How to organize snippets by library for machine learning engineers
Ways for machine learning engineers to organize code snippets by library, with folders, groups and a checklist for comparing tools before you pick one.
SupaKeeper is published by LunarWerx, the makers of AgentHydra, Copilot AI Productivity Suite, Codex Forecast and Model Odds, which are compared here alongside other tools.
To organize snippets by library, make one folder for each library you use, then sort each snippet into the task it supports. Check the list every few weeks so it keeps matching the projects you work on.
| Option | What it can do | What to check |
|---|---|---|
| Copilot AI Productivity Suite | Nested folders and groups for snippets. [1] | Check that your editor is VS Code 1.116.0 or newer, or one of the supported forks. [1] |
| Your editor's own folders or notes | Can keep snippet files in the folders you already use. | Check whether it can group snippets by library and share them with teammates. |
| Another snippet manager | Varies by product, so compare the features listed on its own page. | Check how it groups snippets by library, which editors it supports, and what it costs. |
What you need before you start
Before you organize snippets by library, collect the snippets you already reuse, including the ones scattered across notebooks, scripts and old chat history. Know which libraries your work depends on most, such as pandas, NumPy, PyTorch, scikit-learn or Hugging Face Transformers, so you can name each group clearly.
Decide what a useful snippet looks like in your work. For a machine learning engineer, that might be a data loading pattern, a training loop skeleton, an evaluation metric call or a plotting helper. Keep each snippet small enough to reuse without rewriting it.
Have the editor and coding assistant you use every day ready, because the method should work where you write code. If you work across several machines or share code with teammates, note that before you choose a method, since it affects how the list needs to be shared.
Step by step
- Gather every snippet you reuse into one list, including snippets saved in files, notes and old chats.
- Create one top-level group for each library, such as pandas or PyTorch.
- Inside each library, add subgroups for the task, such as loading data, training, evaluation or plotting.
- Name each snippet with its library and purpose, such as pandas read parquet with explicit dtypes, so it is easy to find later.
- Mark snippets that depend on a specific library version or setting, so you know what to check before reusing them.
- Group snippets that always appear together, so one selection brings in the full set.
- Keep the values that change between projects, such as file paths or learning rates, easy to replace, so the same snippet can be reused without edits.
- Review the list every few weeks, moving out-of-date snippets and removing duplicates.
- Keep a copy of the list where teammates can read it, so the organization is shared rather than held by one person.
Tools that can do it
Copilot AI Productivity Suite is made by the team that publishes this site, SupaKeeper.
Copilot AI Productivity Suite is a VS Code extension that needs VS Code 1.116.0 or newer, and it is also on Open VSX for Cursor, Windsurf and VSCodium. [1] You install it from the VS Code Marketplace as LunarWerx.copilot-suite. [1]
Snippets can sit in nested folders, so each library can have its own folder and each task its own subfolder. You can also bundle several snippets into a group that is referenced all at once, which suits the sets of snippets that always go together. [1]
A snippet can leave a blank that asks you for a value each time you insert it, optionally with a default value. [1] You can select several snippets and move them, delete them or insert all their references at once. [1]
A bundled local MCP server, a standard way for coding tools to connect to outside tools, makes your snippets and groups available to Claude Code, Codex CLI and other MCP clients. [1] If you turn on the slash command export, each snippet is written out as a slash command file for Claude Code and Codex CLI, and the files stay in sync as you edit. [1]
You can export snippets as JSON or make a share link, and teammates can import it. [1] While VS Code telemetry is on, the extension sends anonymous usage statistics, but never snippet content. [1]
The Free plan costs $0 and allows 20 Copilot insertions a week. [1] The Premium plan costs $6.99 a month and has unlimited Copilot insertions. [1]
Other tools and approaches can also organize snippets by library. Check how each one groups snippets, whether it works in the editor you use, whether it can share the list with teammates, and what it costs before you choose.
Common mistakes to avoid
Avoid naming snippets only by task, because a name like training loop does not tell you which library it uses. Include the library in the name so the list stays readable as it grows.
Avoid putting every snippet in one long list. Without groups for each library, searching becomes slow and reusing a snippet means reading through unrelated ones.
Avoid saving snippets that depend on a library version without a note. A pattern that worked with one release may behave differently after an update, so mark version-specific snippets.
Avoid building the list alone if your team shares code. Check whether teammates can read, import or share the list, so each person does not rebuild the same collection.
Avoid skipping the review step. A list that is never cleaned up collects duplicates and old patterns that slow down the work it was meant to speed up.
Frequently asked questions
How do ML engineers organize snippets by library?
Start by creating one folder for each library you use often, then add subfolders for tasks such as loading data, training or evaluation. Name each snippet with its library and purpose so it is easy to find later. Review the folders every few weeks, removing duplicates and out-of-date snippets, so the list stays useful as your projects change.
Can Copilot AI Productivity Suite sort snippets into folders for each library?
Yes, snippets can sit in nested folders, so one folder can hold a library and subfolders can hold its tasks. [1] You can also bundle several snippets into a group that is referenced all at once, which helps when a set of snippets always goes together. [1] Check that the folder layout matches how your projects import each library.
Does Copilot AI Productivity Suite work with Cursor and Windsurf?
Copilot AI Productivity Suite is a VS Code extension that needs VS Code 1.116.0 or newer. [1] It is also on Open VSX for Cursor, Windsurf and VSCodium. [1] You install it from the VS Code Marketplace as LunarWerx.copilot-suite. [1] Confirm your editor is on the supported list before you start.
What does the free plan of Copilot AI Productivity Suite allow?
The Free plan costs $0 and allows 20 Copilot insertions a week. [1] The Premium plan costs $6.99 a month and has unlimited Copilot insertions. [1] Compare those limits with how many snippets you insert in a typical week before you decide which plan fits your work.
Can Copilot AI Productivity Suite share snippets with teammates?
You can export snippets as JSON or make a share link, and teammates can import it. [1] If you turn on the slash command export, each snippet is written out as a slash command file for Claude Code and Codex CLI, and those files stay in sync as you edit. [1] Check that this matches how your team works before you turn it on.
There is a Free plan at $0 that allows 20 Copilot insertions a week. [1]
Try Copilot AI Productivity SuiteSources
Related comparisons
- How to advise customers on AI model upgrades
- How AI community managers answer release date questions
- How to anticipate rate limit interruptions for API integrators
- How to audit agent commands in each session in 2026
- How to audit agent sessions across projects
- Audit What an Agent Ran in Each Session for DevOps
- How freelance coders avoid hitting limits mid-task
- How to avoid hitting the weekly cap for Claude users
- How to avoid running out mid-project in Cursor
- How to budget AI usage per sprint