Initial commit: Crumbforest Architecture Refinement v1 (Clean)

compose/init/03_rag_tracking.sql

@@ -0,0 +1,38 @@
+-- compose/init/03_rag_tracking.sql
+-- RAG tracking tables for indexing status and analytics
+
+-- Track which posts are indexed in Qdrant
+CREATE TABLE IF NOT EXISTS post_vectors (
+  id            INT AUTO_INCREMENT PRIMARY KEY,
+  post_id       INT NOT NULL,
+  collection_name VARCHAR(100) NOT NULL,
+  vector_ids    JSON NOT NULL COMMENT 'Array of Qdrant point IDs for this post',
+  provider      VARCHAR(50) NOT NULL COMMENT 'Provider used for embeddings (openai, openrouter, claude)',
+  model         VARCHAR(100) NOT NULL COMMENT 'Specific model used for embeddings',
+  chunk_count   INT NOT NULL COMMENT 'Number of chunks/vectors created',
+  indexed_at    TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  file_hash     VARCHAR(64) NOT NULL COMMENT 'MD5 hash of post content for change detection',
+
+  FOREIGN KEY (post_id) REFERENCES posts(id) ON DELETE CASCADE,
+  UNIQUE KEY uniq_post_collection (post_id, collection_name),
+  INDEX idx_collection (collection_name),
+  INDEX idx_provider (provider),
+  INDEX idx_indexed_at (indexed_at)
+) ENGINE=InnoDB COMMENT='Tracks RAG indexing status for posts';
+
+-- Track RAG query history for analytics
+CREATE TABLE IF NOT EXISTS rag_queries (
+  id            INT AUTO_INCREMENT PRIMARY KEY,
+  user_id       INT NOT NULL,
+  query_text    TEXT NOT NULL,
+  provider      VARCHAR(50) NOT NULL COMMENT 'Provider used for search/completion',
+  model         VARCHAR(100) NULL COMMENT 'Specific model used',
+  results_count INT NOT NULL COMMENT 'Number of results returned',
+  locale        VARCHAR(8) NULL COMMENT 'Language/locale of the query',
+  created_at    TIMESTAMP NOT NULL DEFAULT CURRENT_TIMESTAMP,
+
+  FOREIGN KEY (user_id) REFERENCES users(id) ON DELETE CASCADE,
+  INDEX idx_user (user_id),
+  INDEX idx_created (created_at),
+  INDEX idx_provider (provider)
+) ENGINE=InnoDB COMMENT='Analytics for RAG queries';