Claude CoWork for Community Managers

What is Claude CoWork?

Claude CoWork is the practice of using Claude as a persistent, context-aware co-worker embedded in your daily community management workflow. This is not about asking a chatbot to draft one post and closing the tab. It is about configuring Claude with your platform, community norms, and member personas so every interaction produces output you can actually use — moderation copy that fits your tone, onboarding sequences that feel human, and stakeholder reports that do not require a full rewrite.

Claude-native prompts. The prompts in this guide use Claude's native XML tag structure (<context>, <instructions>, <format>, <avoid>) for more precise, consistent output. These tags help Claude parse your intent with less ambiguity. They work in ChatGPT too, but are optimized for Claude.

Forrester predicts 90% of communities will integrate AI by 2026. The community managers who pull ahead are not the ones who automate everything — they are the ones who offload the repeatable work (moderation logic, onboarding copy, event reminders) so they can spend their attention on strategy, tone, and the relationships that actually drive retention.

This guide walks you through configuring Claude for community management work, the five workflows where it saves you the most time, and the prompt patterns that keep output on-brand and on-platform.

Install the Community Manager Plugin

This guide works on three Claude surfaces. The plugin is the fastest path on two of them. Pick whichever you use:

If you're on Cowork (desktop or mobile app)

Claude Cowork is Anthropic's agentic workspace — Claude completes work autonomously and returns finished deliverables. The Community Manager plugin packages the workflows below as native skills and slash commands.

1. Open the Cowork plugin directory in your desktop app.
2. Filter by Cowork, search for "Community Manager", and click Install.
3. The plugin's slash commands and ambient skills are now available in any Cowork task.

If you don't see the plugin in the directory yet, install via custom marketplace: paste https://github.com/alexclowe/awesome-claude-cowork-plugins in your Cowork plugin settings.

If you're on Claude Code (CLI)

Install from your terminal:

claude plugin add alexclowe/awesome-claude-cowork-plugins/community-manager

The plugin's slash commands and skills load on next session.

If you're on Claude.ai (web chat only)

Plugins aren't directly installable on the web chat surface. You have two options:

1. Use the prompts in this guide directly in a Claude Project (covered in the next section). Same outputs, more typing.
2. Upload the plugin's skills as a zip via Settings → Features → Custom Skills (Pro/Max/Team/Enterprise plans). Higher friction; only worth it if you want the auto-activating skills, not the slash commands.

What the plugin gives you (any surface)

Slash command	What it does
/moderate-community	Generate moderation playbooks for spam/toxicity/harassment with platform-specific actions and escalation rules
/onboard-member	Draft a 7-day onboarding sequence with welcome messages, FAQ replies, role assignments, and first-interaction prompts
/analyze-sentiment	Aggregate community mood from a sample of messages, flag churn signals, and recommend engagement tactics
/run-event-workflow	Manage event registrations, reminder cadence, post-event follow-up, and participation reporting

Auto-activating skills (no command needed — Claude applies them when relevant):

• Community Health Monitoring — Track member engagement, retention, churn predictors; flag disengagement patterns early
• Moderation Audit Trail — Maintain compliance logs for corporate/regulated communities; document escalation chains

The plugin works standalone for one-off tasks. Pair it with the surface-specific setup below for persistent context across every task — that combination is the full Claude CoWork setup.

Setting Up Claude for Community Management Work

Surface note: The Project setup below is for claude.ai web users. Cowork users have their own task-context mechanism (set context once when starting a Cowork task). Claude Code users get the plugin's ambient skills automatically — no Project setup needed. The workflows themselves are surface-agnostic — paste the prompts wherever you're working. The key to consistent output is using Claude Projects. A Project stores your community context across every conversation so you are not re-explaining your platform or community rules each time.

Step 1: Create a Community Project. In Claude, click "Projects" and create one called something like "Community Management — [Community Name]."

Step 2: Set your custom instructions. In the Project settings, add:

You are my community management assistant. Here is my context:

<community-profile>
- Community name: [Name]
- Platform: [Slack / Discord / Circle / Skool / Discourse / Forum / etc.]
- Community type: [SaaS product / Creator / Course / Professional / B2B / etc.]
- Member count: [Approximate]
- Primary audience: [Role, industry, or interest]
- Brand voice: [Approachable / Authoritative / Casual / Professional]
- Moderation stance: [Link to or summarize community guidelines]
</community-profile>

<rules>
- All moderation language must match the community's established voice and guidelines
- Never disclose member-identifying information in any output
- Flag any content touching COPPA (under-13 users), DSA obligations, or NCMEC-reportable material as requiring immediate human escalation — do not draft a response for those cases
- Outputs are drafts for CM review before publishing or sending
</rules>

Step 3: Upload your community documents. Add your community guidelines, moderation rubric, onboarding checklist, and any platform-specific role or permission structures to the Project knowledge base.

Step 4: Add Feverbee health metrics as context. Paste your L7 (last 7-day active members) and L28 (last 28-day active) baselines, and your first-7-day activation rate. Claude will reference these when generating health reports and churn-risk analysis.

Step 5: Always work inside this Project. Every new conversation inherits your community context automatically.

Five High-Leverage Workflows

1. Moderation Playbook

A written playbook makes moderation consistent across your team and reduces the cognitive load of every judgment call. Claude drafts the full rubric from your platform and policy inputs.

<context>
Platform: Discord. Community: 4,200-member SaaS product community. We have community guidelines covering spam, self-promotion, harassment, and off-topic content. We run three tiers of enforcement: warning, temp mute (24hr), and permanent ban. We are subject to DSA transparency requirements as an EU-accessible platform.
</context>

<instructions>
Draft a moderation playbook that covers:
- Spam and unsolicited promotion (first offense / repeat offense)
- Harassment and personal attacks (severity tiers)
- Off-topic posts (soft redirect vs. removal)
- NSFW or harmful content (immediate escalation criteria)
- Escalation path for NCMEC-reportable content and COPPA violations
- DSA-compliant record-keeping note for each removal category
- Note for financial-services communities: any member post containing investment recommendations, performance claims, or securities references must be escalated for compliance review under FINRA Rule 2210 — do not include a CM response template for those cases
Include decision logic as IF/THEN statements where possible.
</instructions>

<format>
Markdown with H3 headings per violation category. Decision trees as numbered IF/THEN steps. Escalation contacts as bracketed placeholders [Escalation Contact]. Under 800 words.
</format>

<avoid>
Specific platform UI instructions that may change; language that promises zero-tolerance in ways that remove moderator discretion; any real member usernames or cases.
</avoid>

Before Claude: 3-4 hours drafting a playbook from scratch or updating a stale one. After Claude: 15 minutes to input your policy context, 20 minutes to review and refine.

2. 7-Day Onboarding Sequence

First-7-day activation is the single strongest predictor of long-term retention in CMX Hub and Feverbee research. Claude drafts the full sequence — welcome DMs, FAQ replies, role assignment prompts, and nudge messages — from your member persona.

<context>
Platform: Circle. Community: Professional learning community for early-career data analysts, ~800 members. First-7-day activation goal: member posts an introduction, completes the start-here course module, and joins one live session. Current activation rate: 34% (Feverbee L7 baseline).
</context>

<instructions>
Draft a 7-day onboarding sequence:
- Day 0: Welcome DM (sent on join) — warm, direct, 3 clear next steps
- Day 1: Follow-up nudge if no introduction posted — low-pressure, peer-framing
- Day 2: Highlight one community feature or resource relevant to their role
- Day 3: Prompt to join the next live session with specific date/time placeholder
- Day 5: Check-in DM if no activity — offer direct help or FAQ link
- Day 7: Activation summary prompt — celebrate what they have done, name what is still available
Include FAQ reply templates for the 5 most common new-member questions.
</instructions>

<format>
Each day as a numbered section. Message copy in blockquotes. Tone: warm, peer-to-peer, never salesy. Under 700 words total.
</format>

<avoid>
Corporate tone; pressure language; assuming members have completed steps they may not have; real member names or identifying details.
</avoid>

Before Claude: 2-3 hours writing a complete sequence with consistent voice. After Claude: 10 minutes to specify the persona, 20 minutes to review and personalize.

3. Sentiment Analysis and Churn Signals

Pasting a sample of recent member messages into Claude for sentiment analysis gives you a fast read on community health before it shows up in your L28 numbers. Claude flags engagement signals and churn risk patterns.

<context>
I am pasting 30 messages from our community's #general and #support channels over the past 7 days. Community type: B2B SaaS customer community. We track Feverbee health metrics: L7 active = 12% (down from 18% last month), first-7-day activation = 29%.
</context>

<instructions>
Analyze the message sample for:
- Overall sentiment (positive / neutral / negative breakdown, approximate %)
- Recurring themes or pain points in negative sentiment
- Members showing churn signals (repeated frustration, farewell language, unresolved support issues)
- Members showing high engagement worth recognizing or nurturing
- 3 specific tactics to address the dominant negative theme
Provide a Feverbee-style health read: is the community trending toward decline, maintenance, or growth based on this sample?
</instructions>

<format>
Executive summary (3 sentences). Sentiment breakdown table. Churn signals as a bulleted list with anonymized message excerpts. Engagement highlights as a bulleted list. Tactics as numbered recommendations. Under 500 words.
</format>

<avoid>
Identifying real members by username; overfitting conclusions to a small sample without noting the limitation; presenting sentiment percentages as statistically precise.
</avoid>

Before Claude: 45-60 minutes manually reading and tagging a message sample. After Claude: 5 minutes to paste the sample, 10 minutes to review the analysis and act on it.

4. Event Workflow

Community events require four discrete content phases: registration copy, reminder sequence, live facilitation prompts, and post-event follow-up. Claude handles all four from a single brief.

<context>
Event: Monthly community Q&A with a guest expert. Platform: Zoom + Slack recap. Audience: 350-member creator community. Event length: 60 minutes. Guest: [Name, Title, Expertise placeholder]. Registration via a simple RSVP link.
</context>

<instructions>
Generate a complete event workflow:
1. Registration announcement post (Slack/Discord, 150 words max)
2. 48-hour reminder message (email or DM, 100 words)
3. 1-hour-before reminder (brief, high-energy, 50 words)
4. 5 prepared host questions for the live Q&A
5. Post-event follow-up message with recording link placeholder and 1 key takeaway prompt
6. Participation report template: fields for RSVP count, attendee count, peak attendance, top 3 questions asked, engagement notes
</instructions>

<format>
Each phase as a numbered section with copy in blockquotes. Report template as a fillable table. Total under 600 words.
</format>

<avoid>
Overpromising on guest availability or content; copy that duplicates exactly across reminder messages; jargon specific to one platform that would not translate to a recap post.
</avoid>

Before Claude: 90 minutes creating event content across all four phases. After Claude: 10 minutes to brief the event details, 20 minutes to review and schedule.

5. Weekly Community Health Report

A concise stakeholder report keeps leadership aligned without requiring you to mine dashboards every week. Claude structures your metrics into a narrative with context and recommended actions.

<context>
Community: 1,200-member customer community for a project management SaaS. This week's metrics: L7 active members = 210 (17.5%, up from 15.1% last week). L28 active = 380 (31.7%). New members this week: 42. First-7-day activation rate: 38%. Top discussion thread: 180 replies on a feature request for bulk export. 2 support escalations resolved. 1 moderation action (spam removal).
</context>

<instructions>
Write a weekly community health report for a VP of Customer Success audience. Include:
- Health summary: L7, L28, activation rate with trend direction
- Highlights: what drove the spike in activity
- Opportunities: one member segment or thread worth acting on this week
- Risks: any signals worth watching
- Recommended actions: 2-3 specific items for the coming week
Keep it tight — this is an async Slack message, not a slide deck.
</instructions>

<format>
5 short sections with bold headers. Metrics inline, not in tables. Total under 300 words. Tone: confident, direct, lightly conversational.
</format>

<avoid>
Padding with definitions of metrics the audience already knows; hedging every data point; recommending actions that are not actionable within the week.
</avoid>

Before Claude: 30-45 minutes compiling and writing the weekly report. After Claude: 5 minutes to paste metrics, 10 minutes to review and send.

What This Looks Like in Your Week

Monday: Paste last week's message sample into Workflow 3 for a sentiment read before your weekly planning. Use the health report workflow to draft your stakeholder update — send it before your first meeting.

Tuesday/Wednesday: Any new members who joined in the past 7 days and have not activated get the Day 5 or Day 7 nudge from Workflow 2. Takes 10 minutes with Claude drafting the copy.

Thursday: Draft or update your event content for any sessions happening next week using Workflow 4. The full workflow takes 30 minutes instead of 90.

Friday: If a moderation issue came up during the week, use Workflow 1 to update or gap-fill your playbook while the decision is fresh. Five minutes now saves an hour of inconsistency next time.

As needed: Refresh your onboarding sequence every quarter. Paste the current version into Claude and ask: "What is stale, what is missing, and what does not match our current activation data?"

What to Avoid

Posting Claude drafts without voice review. Claude hits your brief but not always your exact tone. Every message that goes to members needs a human read for warmth and platform fit before it publishes.

Using Claude for NCMEC, COPPA, or DSA escalation decisions. These are human and legal escalations. The playbook workflow includes placeholders for escalation contacts — those contacts, not Claude, handle the decision.

Treating sentiment analysis as statistically precise. A 30-message sample gives directional signal, not a statistically valid breakdown. Present findings as "themes from a sample" not percentages in a board deck.

Over-automating onboarding DMs. The goal of the onboarding sequence is to feel personal. If your platform allows it, review each Day 0 welcome DM before it sends and add one line specific to what the member wrote in their application or introduction.

Neglecting to update your Project context. If your community guidelines change, your platform migrates, or your activation baseline shifts meaningfully, update your Project instructions. Stale context produces stale output.

Resources

• Explore the Community Manager AI plugin for ready-to-use prompt workflows built for your platform
• Browse the community manager profession guide for a fuller picture of where AI fits in the role
• Run the AI readiness audit for community managers to identify your highest-leverage starting point