The Claude Code ecosystem: what to take and what to skip
There are now more catalogues and templates around Claude Code than there is time to check them. Here is which repositories actually save hours, and which are just long lists of links.
In this article5
Take three things from the Claude Code ecosystem: one catalogue for the overview, one set of templates matched to your stack, and one measurement tool. Everything else is the same links in a different order. Below is a breakdown of specific repositories, each with the date we checked it.
In eighteen months an ecosystem has grown around Claude Code that is now genuinely hard to navigate. More than fifteen thousand plugins have been indexed, there are dozens of catalogues, and nearly every one calls itself "the most complete". The trouble is that completeness is worth nothing here: a list of four hundred links helps exactly as much as searching GitHub does.
What is actually out there
Four types, kept apart because they solve different problems and confusing them is expensive.
| Type | What it does | When it helps | Risk |
|---|---|---|---|
| Catalogues | Collect links to everything | First survey of a subject | Go stale quietly |
| Templates | Ready-made CLAUDE.md, skills, hooks | Starting a new project | Written for someone else's stack |
| Toolkits | Bundles of agents and commands | A methodology out of the box | Impose someone else's process |
| Measurement | Show you what is happening | Once you are already working | Need their own API keys |
The most common mistake is starting with a toolkit. You install a bundle of a hundred and thirty agents and end up with someone else's methodology laid over your project — and no way to debug it afterwards.
Catalogues
The ecosystem's head catalogue: skills, hooks, slash commands, agent orchestrators, plugins. Maintained by hand, and the quality of the selection shows it. Take it as an entry point — but not as a working tool: it is a list, not an assessment.
A good catalogue answers the question "what exists". None of the existing ones answers "which of these is for me" — because that would mean rechecking every entry, and nobody wants to do that work for free.
How to read any catalogue
Look at the last-commit date on the entries, not on the catalogue. A catalogue can be updated weekly while half the links in it point at repositories abandoned a year ago.
Measurement
An open platform: it shows where ChatGPT, Claude, Gemini, Perplexity and AI Overviews cite you. Self-hosted. Watching rather than taking it right away — at our volume a simpler CLI still covers it, and this one needs setup and API keys of your own.
How to check a repository in five minutes
The order that saves hours. It is also what we run ourselves before recommending anything.
- 1
Last commit date
More than six months of silence in a fast-moving ecosystem almost always means the code was written against older behaviour of the tool.
- 2
Open issues
Not the count — the dates on the replies. A hundred open issues with a live discussion beats three and a silent author.
- 3
The README, for specifics
If there is not a single command or a single example of output in it, the repository was probably assembled from other people's READMEs.
- 4
An actual run
Five minutes to install it and carry out one scenario. That alone rules out more than half of them.
$ git clone --depth 1 https://github.com/owner/repo.git$ cd repo && git log -1 --format='%cr'3 weeks ago$ ls .claude/skills/ hooks/ settings.json
What this means in practice
Create unique, useful content with a point of view of its own, rather than restating what is already on the web.
The same logic applies to repositories. The value is not in the size of the list; it is in someone having actually checked every entry and being willing to stand behind what they said.
Versions move
The star counts and commit dates on the cards above are pulled from GitHub automatically; the notes are written by hand on the date we checked. These tools ship often — if you are reading this months later, check the repository's current state.
Next step
From chat to app: 3 ways to put AI to work
The terminal, the browser and the chat, worked through on a real project — from the first prompt to a published landing page.
Comments