Community Platform Patterns

Community Platform Patterns

Build engaging community platforms with sustainable moderation and meaningful interaction design.

Core Data Model

from datetime import datetime
from enum import Enum

class MemberRole(str, Enum):
    MEMBER = "member"
    MODERATOR = "moderator"
    ADMIN = "admin"

class ContentStatus(str, Enum):
    DRAFT = "draft"
    PUBLISHED = "published"
    HIDDEN = "hidden"      # Soft-removed by mod
    ARCHIVED = "archived"

# Core entities
class Member:
    id: str
    display_name: str
    bio: str
    joined_at: datetime
    role: MemberRole
    reputation: int = 0

class Discussion:
    id: str
    title: str
    body: str
    author_id: str
    category: str
    status: ContentStatus
    created_at: datetime
    reply_count: int = 0
    view_count: int = 0
    pinned: bool = False
    locked: bool = False

class Reply:
    id: str
    discussion_id: str
    author_id: str
    body: str
    parent_id: str | None = None  # Threaded replies
    status: ContentStatus
    created_at: datetime
    upvotes: int = 0

Discussion Forum

Thread Display Patterns

Flat (chronological): Best for announcements and linear conversations.

Threaded (nested): Best for technical discussions where tangents are valuable.

Hybrid (flat with quoted replies): Best for general discussion.

async def get_discussion_with_replies(discussion_id: str, sort: str = "chronological"):
    discussion = await db.get_discussion(discussion_id)
    replies = await db.get_replies(discussion_id)

    if sort == "chronological":
        return sorted(replies, key=lambda r: r.created_at)
    elif sort == "threaded":
        return build_thread_tree(replies)
    elif sort == "popular":
        return sorted(replies, key=lambda r: r.upvotes, reverse=True)

Categories and Tags

CATEGORIES = {
    "reading-group": {"description": "Book discussions and study groups", "color": "#4A90D9"},
    "announcements": {"description": "Official updates", "color": "#E74C3C", "mod_only": True},
    "help": {"description": "Questions and support", "color": "#2ECC71"},
    "showcase": {"description": "Share your work", "color": "#F39C12"},
    "meta": {"description": "About the community itself", "color": "#9B59B6"},
}

Reputation System

Action-Based Scoring

| Action | Points | Rationale | |--------|--------|-----------| | Post discussion | +2 | Encourages contribution | | Reply to discussion | +1 | Encourages engagement | | Receive upvote | +5 | Community validation | | Give upvote | +1 | Encourages voting | | Marked as answer | +15 | High-value contribution | | Post hidden by mod | -10 | Discourages violations |

Privilege Tiers

PRIVILEGE_TIERS = {
    0: ["read", "reply"],
    10: ["create_discussion", "upvote"],
    50: ["edit_own_posts", "flag_content"],
    100: ["create_events", "add_tags"],
    500: ["edit_wiki_posts"],
    1000: ["nominate_moderator"],
}

def get_privileges(reputation: int) -> list[str]:
    privileges = []
    for threshold, perms in PRIVILEGE_TIERS.items():
        if reputation >= threshold:
            privileges.extend(perms)
    return privileges

Content Moderation

Moderation Queue

class ModerationAction(str, Enum):
    APPROVE = "approve"
    HIDE = "hide"
    WARN = "warn"
    BAN_TEMPORARY = "ban_temporary"
    BAN_PERMANENT = "ban_permanent"

async def flag_content(content_id: str, reason: str, reporter_id: str):
    flag = await db.create_flag(
        content_id=content_id,
        reason=reason,
        reporter_id=reporter_id,
    )
    # Auto-hide if threshold reached
    flag_count = await db.count_flags(content_id)
    if flag_count >= 3:
        await db.update_content_status(content_id, ContentStatus.HIDDEN)
        await notify_moderators(content_id, flag_count)

Community Guidelines Enforcement

GUIDELINES = {
    "be_respectful": "Treat others as you'd want to be treated",
    "stay_on_topic": "Keep discussions relevant to the community focus",
    "no_spam": "No promotional content without context",
    "cite_sources": "Back up claims with references when possible",
    "no_personal_attacks": "Criticize ideas, not people",
}

async def moderate(content_id: str, action: ModerationAction, reason: str, mod_id: str):
    await db.record_mod_action(content_id, action, reason, mod_id)
    if action == ModerationAction.HIDE:
        await db.update_content_status(content_id, ContentStatus.HIDDEN)
        await notify_author(content_id, reason)
    elif action == ModerationAction.WARN:
        await send_warning(content_id, reason)

Event Management

class Event:
    id: str
    title: str
    description: str
    organizer_id: str
    event_type: str  # "reading-group", "salon", "workshop", "office-hours"
    starts_at: datetime
    ends_at: datetime
    location: str  # URL for virtual, address for physical
    capacity: int | None
    rsvp_count: int = 0
    recurring: str | None = None  # "weekly", "biweekly", "monthly"

async def create_recurring_events(template: Event, count: int):
    events = []
    for i in range(count):
        event = template.copy()
        if template.recurring == "weekly":
            event.starts_at += timedelta(weeks=i)
            event.ends_at += timedelta(weeks=i)
        events.append(await db.create_event(event))
    return events

Notification System

class NotificationType(str, Enum):
    REPLY = "reply"
    MENTION = "mention"
    UPVOTE = "upvote"
    EVENT_REMINDER = "event_reminder"
    MOD_ACTION = "mod_action"

async def send_notification(user_id: str, notification_type: NotificationType, data: dict):
    prefs = await db.get_notification_preferences(user_id)
    if notification_type.value not in prefs.get("muted", []):
        await db.create_notification(user_id, notification_type, data)
        if prefs.get("email_enabled") and notification_type in prefs.get("email_types", []):
            await send_email_notification(user_id, notification_type, data)

Anti-Patterns

• No onboarding flow — New members need guided first steps
• Flat hierarchy only — Categories/tags help people find relevant content
• No moderation tools — Build moderation from day one, not after the first incident
• Gamification without purpose — Points should encourage healthy behavior, not just activity
• One-size-fits-all notifications — Let members control what they receive
• No archive strategy — Old discussions should be searchable but not clutter active feeds

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: organvm
• Source: organvm/a-i--skills
• License: Apache-2.0
• Homepage: https://organvm.github.io/a-i--skills/

Install and usage instructions live in the source repository linked above.