Semantic Brain Setup

🏗️ Step 1: Create Supabase Project

Go to supabase.com and create a new free project. This will be your brain's backend.

🗄️ Step 2: Get Your Credentials

Navigate to your project's API settings to get the credentials you'll need.

🗃️ Step 3: Enable pgvector Extension

pgvector is a PostgreSQL extension that enables vector similarity search. Without it, your database can't compare embeddings efficiently.

CREATE EXTENSION IF NOT EXISTS vector;

What this does: Installs the pgvector extension if it's not already installed. The IF NOT EXISTS part makes it safe to run multiple times without errors.

📊 Step 4: Create the memories Table

This table stores your thoughts as both text and vector embeddings. Each column has a specific purpose.

CREATE TABLE memories (
  id SERIAL PRIMARY KEY,
  content TEXT NOT NULL,
  embedding vector(1536),
  metadata JSONB DEFAULT '{}'::jsonb,
  created_at TIMESTAMPTZ DEFAULT NOW(),
  importance VARCHAR(20) CHECK (importance IN ('low', 'medium', 'high'))
);

🔧 Step 5: Create the RPC Search Function

This Remote Procedure Call function handles vector similarity search. It takes a query embedding and finds the most similar memories using cosine similarity.

CREATE OR REPLACE FUNCTION search_memories(query_embedding text)
RETURNS TABLE (
  id INTEGER,
  content TEXT,
  metadata JSONB,
  created_at TIMESTAMPTZ,
  importance VARCHAR,
  similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
  RETURN QUERY
  SELECT
    m.id,
    m.content,
    m.metadata,
    m.created_at,
    m.importance,
    1 - (m.embedding::vector <=> query_embedding::vector) as similarity
  FROM memories m
  WHERE m.embedding IS NOT NULL
  ORDER BY m.embedding::vector <=> query_embedding::vector
  LIMIT 10;
END;
$$;

What this does: Takes your query embedding, compares it to all stored memories using cosine distance, and returns the top 10 matches sorted by similarity.

⚙️ Step 6: Configure Your Environment

Create a .env file in your brain skill directory with these credentials.

SUPABASE_URL=https://your-project.supabase.co
SUPABASE_SERVICE_KEY=eyJ...your-service-role-key...
OPENAI_API_KEY=sk-proj-...your-openai-key...

Security note: Never commit .env files to git. Add .env to your .gitignore file.