Social Media Platform Builder#

Build a complete social media management system with plugin architecture, AI-powered content generation, and multi-platform publishing.

Architecture#

text

┌─────────────────────────────────────┐
│         Frontend (5 pages)          │
│  Dashboard│Compose│Calendar│Analytics│Settings│
├─────────────────────────────────────┤
│         API Layer (FastAPI)         │
│  Posts CRUD│Publishing│Calendar│AI  │
├─────────────────────────────────────┤
│      Plugin Registry (per-platform) │
│  Twitter│Instagram│YouTube│FB│TikTok│Manual│
├─────────────────────────────────────┤
│    LangGraph Content Pipeline       │
│  Voice→Research→Draft→Optimize→Save │
├─────────────────────────────────────┤
│         Supabase (6 tables)         │
└─────────────────────────────────────┘

Step 1: Supabase Tables#

Create 6 tables:

• social_posts — id, platform, content, status (draft/scheduled/published/failed), media_urls, published_at, post_url, engagement_metrics (JSONB)
• platform_connections — id, platform, account_name, credentials (JSONB), status, scopes
• content_calendar — id, post_id (FK), scheduled_for, platform, status
• brand_voices — id, name, description, tone, example_phrases (JSONB array), is_default
• social_analytics — id, post_id (FK), platform, impressions, clicks, likes, shares, comments, fetched_at
• publish_queue — id, post_id (FK), platform, status, retry_count, error_message

Seed 3-5 brand voices. Example voices:

• Primary Voice (default) — authoritative, clear, educational
• Casual — conversational, friendly, uses contractions
• Coach — motivational, direct, action-oriented

Step 2: Plugin System#

Base class pattern:

python

class SocialPlugin:
    platform: str
    def validate_credentials(self, creds: dict) -> bool
    def publish(self, content: str, media_urls: list = None) -> dict
    def get_analytics(self, post_id: str) -> dict
    def format_content(self, content: str, max_length: int) -> str

class PluginRegistry:
    _plugins: dict[str, SocialPlugin] = {}
    def register(self, plugin: SocialPlugin)
    def get(self, platform: str) -> SocialPlugin
    def list_active(self) -> list[str]

Platform-specific implementations:

• Twitter: tweepy or requests to v2 API. 280 char limit. Free tier = 100 posts/month.
• Facebook: Graph API v21.0. Page Access Token required. Post to /{page_id}/feed.
• Instagram: Graph API via FB Page Token. Post to /{ig_user_id}/media → /{ig_user_id}/media_publish. Image required.
• YouTube: google-auth + google-api-python-client. OAuth with youtube.upload scope. Upload via resumable upload API.
• TikTok: Content Posting API (requires app review + demo video). Use Manual mode as fallback.
• Manual: No API — generates copy + suggests optimal posting times. Fallback for any unconnected platform.

Step 3: LangGraph Content Pipeline#

6-node graph:

text

load_voice → research_context → generate_drafts → optimize_per_platform → finalize → END

• load_voice: Pull selected brand voice from Supabase
• research_context: Query knowledge vault / RAG for relevant domain content
• generate_drafts: LLM generates 2-3 draft variants in the voice
• optimize_per_platform: Adapt each draft per platform (length, hashtags, media suggestions)
• finalize: Store drafts in social_posts as draft status

Step 4: API Endpoints#

Core routes (~19 endpoints):

• GET/POST /api/social/posts — CRUD
• POST /api/social/posts/{id}/publish — publish to selected platforms
• GET/POST /api/social/calendar — calendar view + scheduling
• GET /api/social/analytics — aggregated metrics
• GET/POST /api/social/voices — brand voice management
• GET/POST /api/social/connections — platform credentials
• POST /api/social/generate — AI content generation (triggers LangGraph)

Step 5: Frontend Pages#

5 pages with shared dark-theme shell (sidebar nav, top bar):

1. Dashboard — post count by platform, recent activity, quick-publish
2. Compose — rich editor, platform multi-select, voice picker, live preview cards per platform
3. Calendar — month view, color-coded by platform, drag-to-reschedule
4. Analytics — Chart.js line/bar charts for engagement over time
5. Settings — platform connection forms (OAuth flows), brand voice editor

Key Patterns#

• Credentials in JSONB: Each platform stores different auth shapes (API keys vs OAuth tokens vs page tokens) in a single credentials JSONB column
• Graceful degradation: Always include Manual plugin as fallback — generates copy even without API access
• Platform content limits: Enforce per-platform (Twitter 280, IG needs image, YT needs video, FB 63K)
• Publish queue with retry: Failed publishes go to retry queue with exponential backoff