/learn:setup

Category: Learning Purpose: Set up the Supabase learning system database schema

Usage

From Claude Code terminal:

./scripts/supabase/setup.sh           # Run setup
./scripts/supabase/setup.sh --check   # Check if schema exists
./scripts/supabase/setup.sh --help    # Show help

Or via command:

/learn:setup                    # Interactive setup
/learn:setup --check            # Check if schema exists

CRITICAL: Claude MUST Execute

When user runs /learn:setup, Claude MUST actually run the setup script or verify schema.

DO NOT just show instructions. DO check if schema exists and run setup if needed.

# ALWAYS source env first
source .envrc 2>/dev/null || source .claude/.envrc 2>/dev/null || true

Setup Flow

Step 1: Bootstrap (One-time)

Run bootstrap.sql in Supabase SQL Editor:

File: scripts/supabase/bootstrap.sql

CREATE OR REPLACE FUNCTION exec_sql(query text)
RETURNS json LANGUAGE plpgsql SECURITY DEFINER AS $$
BEGIN
  EXECUTE query;
  RETURN json_build_object('success', true);
EXCEPTION WHEN OTHERS THEN
  RETURN json_build_object('success', false, 'error', SQLERRM);
END;
$$;

Step 2: Run Setup Script

./scripts/supabase/setup.sh

This will:

Check environment variables
Verify exec_sql function exists
Create all tables, views, and functions
Show progress and any errors

Alternative: Manual Setup

Copy and paste scripts/supabase/schema.sql into Supabase SQL Editor.

Execution Steps

When user runs /learn:setup, Claude should:

Step 1: Check Environment Variables

echo "Checking environment..."
echo "SUPABASE_URL: ${SUPABASE_URL:+✓ set}"
echo "SUPABASE_SECRET_KEY: ${SUPABASE_SECRET_KEY:+✓ set}"

If either is missing, tell user to:

Copy .envrc.template to .envrc
Set the Supabase credentials
Run source .envrc

Step 2: Check if Schema Already Exists

# Test if af_feedback table exists
response=$(curl -s -w "%{http_code}" -o /tmp/schema_check.json \
  "${SUPABASE_URL}/rest/v1/af_feedback?limit=1" \
  -H "apikey: ${SUPABASE_SECRET_KEY}" \
  -H "Authorization: Bearer ${SUPABASE_SECRET_KEY}")

if [ "$response" = "200" ]; then
  echo "✅ Schema already exists"
  exit 0
fi

Step 3: Check if Bootstrap Function Exists

# Test if exec_sql function exists
response=$(curl -s -w "%{http_code}" -o /tmp/bootstrap_check.json \
  -X POST "${SUPABASE_URL}/rest/v1/rpc/exec_sql" \
  -H "apikey: ${SUPABASE_SECRET_KEY}" \
  -H "Authorization: Bearer ${SUPABASE_SECRET_KEY}" \
  -H "Content-Type: application/json" \
  -d '{"query": "SELECT 1"}')

if [ "$response" != "200" ]; then
  echo "❌ Bootstrap function not found"
  echo ""
  echo "Run this SQL in Supabase Dashboard → SQL Editor:"
  echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
  cat scripts/supabase/bootstrap.sql
  echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
  exit 1
fi

Step 4: Run Setup Script

./scripts/supabase/setup.sh

Or if running manually:

node scripts/supabase/setup-schema.cjs

Step 5: Verify Setup

# Confirm tables were created
curl -s "${SUPABASE_URL}/rest/v1/af_feedback?limit=1" \
  -H "apikey: ${SUPABASE_SECRET_KEY}" \
  -H "Authorization: Bearer ${SUPABASE_SECRET_KEY}"

echo "✅ Learning system setup complete!"
echo "Run /learn:status to verify"

Output

🐸 Aura Frog - Supabase Schema Setup

📡 Supabase URL: https://your-project.supabase.co

1. Checking exec_sql function...
   ✓ exec_sql function exists

2. Reading schema file...
   ✓ Read 15234 bytes

3. Parsing SQL statements...
   ✓ Found 42 statements

4. Executing statements...
   [1/42] Executing... ✓
   [2/42] Executing... ✓
   ...
   [42/42] Executing... ✓

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📊 Summary:
   ✓ Successful: 42

✅ Schema setup complete!

Next steps:
   1. Run /learn:status to verify connection
   2. Run /learn:analyze to check for patterns

Tables Created

Table	Purpose
`af_feedback`	User corrections, approvals, ratings
`af_workflow_metrics`	Workflow execution data
`af_agent_performance`	Agent success tracking
`af_learned_patterns`	Identified patterns
`af_knowledge_base`	Persistent learnings

Views Created

View	Purpose
`v_agent_success_rates`	Agent performance by task
`v_common_patterns`	Pattern summary
`v_improvement_suggestions`	Actionable suggestions
`v_workflow_trends`	Weekly trends
`v_feedback_summary`	Feedback statistics

Functions Created

Function	Purpose
`record_feedback()`	Insert feedback record
`update_pattern_frequency()`	Increment pattern count
`get_agent_recommendation()`	Get best agent for task

Troubleshooting

"exec_sql function does not exist"

Run this in Supabase SQL Editor first:

CREATE OR REPLACE FUNCTION exec_sql(query text)
RETURNS json LANGUAGE plpgsql SECURITY DEFINER AS $$
BEGIN
  EXECUTE query;
  RETURN json_build_object('success', true);
EXCEPTION WHEN OTHERS THEN
  RETURN json_build_object('success', false, 'error', SQLERRM);
END;
$$;

"Permission denied"

Verify SUPABASE_SECRET_KEY is the Secret key (not Publishable)
Check RLS policies allow service role access

"Table already exists"

Safe to ignore - the schema uses CREATE TABLE IF NOT EXISTS.

Related Commands

/learn:status - Check learning system status
/learn:analyze - Run pattern analysis
/learn:apply - Apply improvements

Version: 1.0.0

/setup