api-database-surrealdb

# SurrealDB Patterns > **Quick Guide:** Use the `surrealdb` SDK (v2+) with `new Surreal()` and `connect()`. Model relationships with record links for simple pointers and `RELATE` for graph edges with metadata. Use `SCHEMAFULL` tables in production with `DEFINE FIELD` constraints. Always use parameterized queries (`$variable`) to prevent injection. Record IDs are `table:id` -- they are immutable and first-class values in SurrealQL. Live queries push changes without polling. --- <critical_requirements> ## CRITICAL: Before Using This Skill > **All code must follow project conventions in CLAUDE.md** (kebab-case, named exports, import ordering, `import type`, named constants) **(You MUST use parameterized queries with `$variables` for ALL user input -- string interpolation in SurrealQL enables injection attacks)** **(You MUST use `new RecordId("table", "id")` in SDK v2 -- plain `"table:id"` strings are NOT automatically parsed as record IDs)** **(You MUST call `db.use({ namespace, database })` or pass namespace/database in `connect()` options BEFORE any queries -- queries without a selected namespace/database silently fail or error)** **(You MUST NOT rely on `SCHEMALESS` tables in production -- use `SCHEMAFULL` with `DEFINE FIELD` to enforce data integrity at the database layer)** **(You MUST NOT use `UPDATE`/`DELETE` with `WHERE` on large tables without indexes -- SurrealDB currently does not use indexes for UPDATE/DELETE WHERE clauses (use subquery workaround))** </cr...