Community Manager Toolkit

A free plugin for Claude and ChatGPT that helps community managers introduce members to each other, match people at events, welcome and re-engage members, and prove community ROI with outcomes instead of activity.

Built from what worked in a real ~850-member professional network. Free and open source (MIT). No account, no server, nothing collected.

Five skills for community management

Ask in plain words: “Match the attendees of my next Luma event”, “Help me introduce two members, double opt-in” or “Build this quarter’s community impact report”.

Why it works this way

The rules come from one ~850-member professional network running member introductions over WhatsApp from June to September 2026. Figures are rounded, come from that one network, and are a reference point, not a benchmark.

• A named person plus a yes/no beats a feature tour. “[Name] asked to meet you, yes or no?” got replies about 63% of the time. A generic event reminder got about 28% and a one-off broadcast to members never contacted before about 18%.
• Roughly 1 intro request in 3 became a real conversation, and when people did meet they rated it about 4.7 out of 5. You only know that if you log every introduction and ask afterwards.
• New members who were introduced to a named person in their first days replied about 59% of the time, against about 47% for a plain welcome message.
• Registration is not attendance. At one event only about 30% of pre-matched pairs had both people turn up, so every guest gets a backup.

Guardrails built in

• Drafts only. Nothing is sent unless you explicitly ask, one message at a time. Nothing is ever bulk-sent.
• Consent first. Nobody’s details go to another person until they have said yes.
• No same-organisation pairs, and identity checks (full name and organisation) when two members have similar names.
• Professional reasons only. Match reasons never mention or infer sensitive attributes.
• Data minimisation. Reports and the intro ledger never include phone numbers or email addresses.
• Cooldowns and spacing so members are not over-messaged, and opt-out or removal requests are honoured immediately.

How to install

The toolkit is a set of instructions (skills) that your AI assistant follows. Pick your assistant:

Claude (claude.ai, Claude Desktop, Claude Code)

1. Open the Claude plugin directory. (Directory link coming once the listing is approved.)
2. Search for “Community Manager Toolkit” and install it.
3. Optionally connect Luma, Notion, HubSpot, Slack or Google Drive from your connector settings.
4. Type a request such as “Match the attendees of my next Luma event”.

ChatGPT

1. Open the ChatGPT Plugin Directory. (Directory link coming once the listing is approved.)
2. Search for “Community Manager Toolkit” and add it.
3. Optionally connect the apps you use, then start with one of the suggested prompts.

Without installing

The skills are plain Markdown files in the plugin/skills folder on GitHub. You can read them, adapt them, or paste one into any AI assistant.

Works with your tools, or just a file

Everything is optional. The toolkit uses whatever you have connected, such as Luma, HubSpot, Notion, Slack, Google Drive, Airtable, Gmail or Circle. With nothing connected, upload a CSV export (including Eventbrite attendee reports) or a WhatsApp chat export.

Privacy in short

• The plugin is instructions only. It has no server and collects and stores nothing.
• Your data is processed by the AI assistant you run it in, and by any tools you connect, under those providers’ terms.
• This website uses no analytics and sets no cookies of its own. It is hosted on GitHub Pages, which has its own logging.
• You are responsible for having a lawful basis and any consent needed for the member data you use (UK GDPR and GDPR).

Read the privacy policy · Terms of use · Support

Questions