AI Guides › Step-by-step guides
By Nigel Guy · 6 min read
The usual mistake with a new Claude account is to collect: a screenshot of "thirty things to install" goes round, you connect a dozen apps in one evening, and a week later Claude is slower to steer and you cannot say what any of it is doing. Directories change often, so a fixed list of thirty goes stale quickly. What lasts is the order you add things in, and checking each one against the official directory rather than a picture of it.
The rule: add things in three layers, built-in first, then connectors, then plugins, and switch each one on only when you can name the task it serves.
I could not verify the "thirty things" list itself, so this guide does not reproduce it. It uses what the official Anthropic documentation currently describes, and tells you where to check the live catalogue.
Use these for a week before moving on. Most people find the built-in layer covers more than they expected.
A connector lets Claude read from, and sometimes write to, an app. Anthropic says connectors respect your existing permissions in the source system: if you cannot see something there, Claude cannot either.
Types worth knowing: web connectors work in Claude, Cowork, Desktop and mobile; desktop extensions work in Claude Desktop only; custom connectors use remote MCP servers, and free users are limited to one.
Pick connectors for the three weekly tasks you wrote down: your email and calendar, your document store, your project tracker. The directory lists names such as Asana, Notion, Linear and Atlassian, but the catalogue moves, so treat the live directory as the source, not this paragraph.
On Team and Enterprise, an Owner must enable connectors for the organisation first; each person then signs in separately.
A plugin bundles skills, connectors, commands and agents so you add them as one unit.
In claude.ai or the desktop app: go to Customize > Plugins. Use Discover to browse, open a plugin to read what it contains, then select Add. The directory's plugins appear on Pro, Max, Team and Enterprise. A plugin does not sign you in to anything: open its Connectors tab to see each connector as Connected, Not connected or Not added. Plugins you add are saved to your account and also reach Cowork and Claude Code.
In Claude Code: the official marketplace, claude-plugins-official, is added automatically the first time you start an interactive terminal session. Then:
/plugin install commit-commands@claude-plugins-official
This opens the plugin's details rather than installing at once. Read the "Will install" and context-cost figures, then choose a scope: user (every project on this machine), project (everyone on this repository) or local (this repository, you only). Confirm it worked by typing "/" and looking for /commit-commands:commit, or run claude plugin list in your shell. Run /plugin with no name to browse the Discover tab.
Anthropic names these as its own: commit-commands, code-review, feature-dev, and the language server plugins. Most of the official marketplace comes from partners and other authors, and the catalogue changes often, so check the Discover tab or claude.com/marketplace/plugins.
For each thing you added, run one of your three real tasks and see whether it did the job better than before. Then check the cost side: in Claude Code, claude plugin details <name> shows what a plugin adds to every session, and the Installed tab marks plugins under "Not used recently". If a plugin did nothing for you in a fortnight, disable or remove it (Disable plugin toggle, or Remove from its menu, on claude.ai).
anthropics/claude-code adds the demo catalogue, not the official one. Where plugins appear in both, install from claude-plugins-official so you do not end up with two copies.