supabase-integration

Supabase Integration

Overview

Centralized database layer for all Shapira ecosystem data persistence. Executes via GitHub Actions workflows - never local curl.

Connection Details

• Project URL: https://mocerqjnksmhcjzxrewo.supabase.co
• Service Role Key: Stored in GitHub secrets as SUPABASE_SERVICE_ROLE_KEY
• Key IAT: 1764532526

Table Schema

BidDeed.AI Tables

Life OS Tables

Shared Tables

Workflow: Insert Data

Trigger: User says "log this", task completion, auction analysis, learning capture

1. Identify target table from context
2. Format payload matching schema
3. Execute via GitHub Actions workflow

# GitHub Actions dispatch (not local curl)
# See: .github/workflows/insert_insight.yml in both repos

Insert Patterns by Domain

BidDeed.AI - Auction Result:

{
  "table": "auction_results",
  "data": {
    "case_number": "05-2024-CA-012345",
    "property_address": "123 Main St, Melbourne, FL",
    "judgment_amount": 150000,
    "max_bid": 82500,
    "recommendation": "BID",
    "liens": [{"type": "HOA", "amount": 5000}]
  }
}

Life OS - Task:

{
  "table": "tasks",
  "data": {
    "description": "Review Q4 insurance renewals",
    "state": "INITIATED",
    "complexity": 6,
    "clarity": 8,
    "domain": "BUSINESS",
    "estimated_minutes": 45
  }
}

Life OS - Learning Session:

{
  "table": "learning_sessions",
  "data": {
    "source": "YouTube: How to 10x Your Real Estate Portfolio",
    "key_takeaways": ["Leverage OPM", "Focus on cash flow"],
    "business_application": "Apply to Everest Capital deal structuring"
  }
}

Workflow: Query Data

Trigger: "show me", "what were", "find", historical lookup

1. Construct query with filters
2. Execute via Supabase client
3. Return formatted results

Query Examples

-- Recent tasks by domain
SELECT * FROM tasks 
WHERE domain = 'BUSINESS' 
AND state != 'COMPLETED'
ORDER BY created_at DESC LIMIT 10;

-- Auction performance
SELECT recommendation, COUNT(*) 
FROM auction_results 
GROUP BY recommendation;

-- Michael's best times by event
SELECT event, MIN(time) as pr 
FROM michael_swim_times 
GROUP BY event;

LangGraph Integration

Input Schema

class SupabaseInput(TypedDict):
    operation: str  # "insert" | "query" | "update"
    table: str
    data: dict  # For insert/update
    filters: dict  # For query

Output Schema

class SupabaseOutput(TypedDict):
    success: bool
    rows_affected: int
    data: list  # Query results
    error: Optional[str]

Node Implementation

def supabase_node(state: dict) -> dict:
    """Execute Supabase operation as LangGraph node."""
    from supabase import create_client
    
    client = create_client(
        "https://mocerqjnksmhcjzxrewo.supabase.co",
        os.environ["SUPABASE_SERVICE_ROLE_KEY"]
    )
    
    if state["operation"] == "insert":
        result = client.table(state["table"]).insert(state["data"]).execute()
    elif state["operation"] == "query":
        query = client.table(state["table"]).select("*")
        for k, v in state["filters"].items():
            query = query.eq(k, v)
        result = query.execute()
    
    return {"success": True, "data": result.data}

Scripts

• scripts/insert_insight.py - Generic insert with validation
• scripts/query_builder.py - Dynamic query construction

References

• references/schema_details.md - Full column definitions and constraints
• references/migration_history.md - Schema evolution log

Best Practices

1. Always use GitHub Actions - Never local curl to Supabase
2. Validate before insert - Check required fields
3. Use timestamps - All tables have created_at/updated_at
4. Category consistency - Use defined categories (BUSINESS, MICHAEL, FAMILY, PERSONAL)