Use community-marketing to start with a small founding group or repair a community where people join without participating. It emphasizes reasons to return, peer help, and responsibility for maintaining discussion rather than treating member count as success.
Install and invoke
Run this in your project terminal to install only this skill for Claude Code:
npx skills add coreyhaines31/marketingskills --skill community-marketing -a claude-code
Node.js and npx must be available. For Codex, replace -a claude-code with -a codex. In the agent conversation, explicitly ask it to use community-marketing and supply your material. The installation command belongs in the terminal, not in your content brief.
Choose a member group and purpose
Explain who should join and what they gain from one another. Product support, peer networking, learning, and product development are different purposes. An early community does not need many channels attempting all of them.
Example brief:
Use community-marketing to plan the first month for a community of design-studio owners.
Twenty users are interested and management capacity is three hours weekly. The goal is exchanging project-handoff experience.
Provide minimal channels, a welcome post, first-week participation, and one recurring discussion.
Give each activity a member benefit and response owner. Do not schedule daily livestreams or automated recruitment messages.
Choose a platform for the interaction required. Live chat, searchable long-form knowledge, and course activities need different capabilities. Check member willingness, information retrieval, and response capacity before choosing a fashionable platform or proposing a migration.
Make the first contribution clear
A welcome path should explain the discussion scope, show how to ask a useful question, and provide examples. Staff can seed conversations transparently, but should not fabricate member activity. One weekly discussion about a concrete challenge is more maintainable than many empty channels.
Plan how to handle unanswered questions, repeated issues, and irrelevant promotion. Turn useful answers into searchable resources with appropriate member permission. When feedback informs a product change, return to the discussion and explain what happened. This gives participation an observable result.
Review initial participation, reply rates, return visits, and non-staff contributions each week. Quiet reading can still be valuable, and a few prolific contributors do not establish universal satisfaction. Supplement metrics with brief feedback. If management capacity is inadequate, reduce activity commitments and preserve basic responsiveness instead of manufacturing messages to increase activity counts.
Draft a first-week community playbook
Do not begin with a complete annual calendar. Choose one member type and one problem members might help each other solve. Ask the skill for five small artifacts: a welcome note, a question people can answer, a self-introduction prompt, an example moderator reply, and the first gathering. These should invite members to share their own experience rather than wait for company broadcasts. Specify which issues belong in public discussion and which should go to private support.
If the community already exists, sort a recent sample of threads into member-initiated, staff-initiated, answered, and unanswered. A high comment count alone does not show that members help one another. If only staff create posts, repair response practices before recruiting more members. The official skill suggests tracking first-week posting, thread replies, and the share of content created by non-staff. Interpret these against the community’s size and purpose instead of treating a single percentage as a universal pass mark.
After a week, ask: Did a newcomer know what to post first? Were questions answered? Did members help one another? Revise the welcome flow and channel layout using those answers. A small community needs human moderation; calling it an automatic growth loop will not make one appear.
Example: repair an established community with unanswered questions
A community with two hundred members may need better replies, not more channels. Suppose thirty help threads appeared over two weeks and half received no answer. Export thread time, topic, first-response time, and responder role, remove private details, then ask: “Group unanswered questions by subject and difficulty. Design a moderator rota, three pinned prompts, and one weekly member-help ritual. Fix the response path before adding recruitment work.”
Check that every category has an appropriate route: can members help one another, must an account-specific issue move to private support, or is the request outside the group’s scope? If the proposal says only “increase engagement” without naming who responds when, ask for a rota someone can actually follow. After a week, inspect unanswered count, first-response time, and the share of replies from members rather than staff. Automated acknowledgements are not member-to-member help.
Match the operating rhythm to the member count
With fewer than one hundred active members, decide who starts discussions, welcomes newcomers, and follows up on unanswered posts each week. More empty channels will not create conversations. As membership grows, add moderation shifts, recurring events, and knowledge curation. Notion’s community team recruited ambassadors from people already sharing templates and experience, then supported their events and contributions. Use that as an operating example: find willing contributors before designing more platform features.
Sush Padhye’s retail adaptation first asks you to choose a community model: VIP loyalty, ambassadors, user-generated content, peer support or a category/lifestyle group. Each has a different member benefit and success signal, and wholesale trade accounts are deliberately outside the usual community model. Start with why members would come for one another before choosing channels and events. If the relationship is account management, do not force public-post metrics onto it. This is an independent method adaptation, not an observed community result.