PROD. SECOND UNIT UGC/SIDES — MODULE 2/DRAFT/Fact-check before publish

Module 2  ·  section 7  ·  1,550 words · ~10 min read

Connectors and plugins — wiring up the tools you already use

assetDoc type
draftStatus
RequiredFact-check
2026-08-14Last verified
Document controlVerified against product docs
Last verified
2026-08-14
Method
support.claude.com — Use connectors to extend Claude's capabilities; Use Google Workspace connectors; Use plugins in Claude

05-tool-landscape.md covers the tools that make video. This covers the plumbing between them and the place you write — which sounds like the boring half and is actually where a working week gets shorter.

Explain-on-first-use: a connector lets Claude reach into an app you already use — read a document, search a drive, put a file somewhere — instead of you copying and pasting between windows. A plugin is a bundle: skills, connectors and other pieces packaged so you install one thing rather than five.

1The honest version of this section

Most "connect everything" advice is written for people whose job is software. A UGC creator needs about three connectors and probably has two of them already. This document names those, says plainly what each is actually for in this work, and then stops — rather than listing forty integrations you will never open.

If a connector is not on this page it is not because it is bad. It is because a creator writing scripts and cutting video has no use for it, and padding this list would waste the one thing this document is for.

2Before you connect anything

Two things, both from Anthropic's own documentation, both worth reading twice:

"Only connect services you trust and need for your workflows."

Third-party services "operate under their own terms on their own infrastructure," which may sit outside the U.S. Every connector you add widens what an assistant can reach.

Connectors inherit your permissions

If you cannot access something in the source system, Claude cannot either — and "even when you allow a write action in Claude, a person still needs the underlying permission in the source system to make that change."

That second one matters more than it sounds for client work: a connector to a drive a client shared with you can reach everything on that drive that you can reach. Before connecting a shared client drive, know what else is on it.

3Adding a connector — the actual steps

Verified 2026-08-14. Expect menu labels to have moved by the time you read this; the shape will be the same.

  1. In a chat, click the + at the lower left, or type /
  2. Hover Connectors, choose Manage connectors
  3. Click the + next to Connectors
  4. Browse by category or search
  5. Click the one you want, read what it can actually do — each directory entry lists its read/write capabilities
  6. Click Connect or Install
  7. Complete authentication and configure settings

The full directory is at claude.ai/connectors, or via Customize at claude.ai/customize/connectors. Browse it once, properly, before you decide what you need — the capability lists tell you more than any tutorial.

On Team or Enterprise plans, an Owner has to enable connectors at the organisation level first, and you still authenticate individually afterwards. If a connector is missing and you are on a team plan, that is usually why.

4Plan check, in one line

All three connectors below work on a free Claude account. Google Workspace connectors — Drive, Gmail, Calendar — are available to all users on Claude and Claude Desktop, verified 2026-08-14. Nothing in the next section requires an upgrade.

Two things later on this page do, and both are marked where they appear: custom connectors (free gets one, paid more) and plugins (paid plans only). Neither is something a UGC creator needs in order to start.

5The three worth having

1. Google Drive — the one that earns its place immediately

Available to all users, including free, on Claude and Claude Desktop. Alongside Gmail and Google Calendar as the Google Workspace connectors.

What it does for this work, verified: Claude can read Google Docs directly, pull data from Sheets (exported as CSV), extract content from Slides (as plain text), and search across your Drive.

Why a creator cares, concretely:

  • The brief arrives as a Google Doc. It almost always does. Reading it in place beats copy-pasting a formatted document and losing the structure.
  • Your rate card and outreach tracker from the Module 0A kit live in Sheets. Pulling those in is how a rate question gets answered from your actual numbers rather than from memory.
  • Client folders. Searching a shared drive for the last approved script beats scrolling it.

One documented reassurance worth knowing, since clients ask: Anthropic states it does not train its models on Gmail, Drive or Calendar connector data.

2. Whatever holds your written work

If your scripts and intakes live somewhere other than Drive, the connector for that place is your second one. The point is not the specific product — it is that your BV-01 intakes and experience files should be reachable without you pasting them, because a file you have to paste is a file you will eventually not paste.

Check the directory for what you use. If there is no connector for it, that is a real argument for moving those particular files somewhere there is one.

3. A calendar — only if you run deliverable deadlines through one

BV-4 records turnaround and revision rounds per client. If your deadlines live in a calendar, connecting it makes "what's due this week and for whom" answerable in one line. If they live in your head or in a notes app, skip this — a connector to an empty calendar is worse than no connector, because it produces confident answers about nothing.

6What not to bother with, and why that's the useful part

  • Developer tooling — issue trackers, code repositories, deployment services. Real connectors, genuinely useful, not for this work.
  • Anything you check less than weekly. A connector to a system you rarely open adds access surface and returns stale context.
  • Connectors "for completeness." Every one you add is another third party with its own terms. There is no prize for having the most.

7Custom connectors — the honest answer

Custom connectors use remote MCP (Model Context Protocol) to reach your own servers. Free users get one; paid plans allow more.

You almost certainly do not need this. It exists for people connecting in-house systems. If you find yourself reading about MCP because a tutorial told you to, stop — you have wandered into developer territory and there is nothing for a creator there. Come back if you ever run a piece of infrastructure that needs it, which for most creators is never.

8Plugins

Plugins require a paid plan

— Pro, Max, Team or Enterprise. Verified 2026-08-14.

Nothing structural is missing on free. A plugin is a convenience wrapper: it bundles things you can otherwise add one at a time. The two Skills that ship with this module are Skills, not plugins — they install by copying a folder, on any plan. Read this section so you know what the word means when you meet it, then carry on.

A plugin bundles skills, connectors and other pieces into one install. Install via Customize → Plugins → Browse plugins → Install. Skills that arrive in a plugin then appear when you type / or click +.

Two things the documentation says that you should not skim past:

"Plugins may include local MCP servers that run on your computer with the same permissions as any other program you run. Only install plugins from sources you trust."

Read that as: installing a plugin is closer to installing software than to ticking a setting. Apply the same standard you would to any download.

"Hooks and sub-agents run only in Cowork, so they appear grayed out in chat."

If part of a plugin looks disabled, that is usually why — those pieces need the agentic surface (06-agentic-primer.md), which is also paid-only.

Anthropic states that plugins with malicious content are blocked automatically where skill scanning is on. Useful, and not a substitute for the trust rule above.

9Where this leaves the two Skills in this module

script-from-intake and claim-guard are Skills, not plugins — folders with a SKILL.md, copied into .claude/skills/. Nothing to install, no plan gate, no MCP server. They work the same on free as on Pro. See skills/README.md.

The connector question for them is narrower and worth stating: if your intake and experience files live in Drive, a Drive connector means the skills can reach them without you pasting. That is the entire integration, and it is enough.

10Setup, end to end

Twenty minutes, once.

  1. Open claude.ai/connectors. Read the directory. Do not install anything yet.
  2. Connect Google Drive (or your equivalent).
  3. Put one real client's intake.md and experience.md where that connector can see them.
  4. Start a chat and ask it to read the intake and summarise the guardrails back to you. If the summary is wrong, your intake is unclear — that is the finding, and it is a useful one.
  5. Stop. Do not connect anything else until something specific annoys you.

That last step is the actual advice. Connectors added because a tutorial listed them sit unused, widen access, and make the next person to audit your setup ask why a video creator's assistant can reach a payments system.

Your word for it — nothing is tracked automatically.

Source: content/module-2/07-connectors-and-plugins.md