Embedded database operations

Check the installed database version, connection lifecycle and filesystem before applying recipes. The relevant evidence is the behavior of the actual configured connections.

Work from the affected contract

1. Identify workload, database build/version, connection ownership, storage/filesystem and durability requirement. Confirm the workload fits the embedded engine.
2. Inspect effective connection/file PRAGMAs and transaction ownership. Configure connection-scoped settings on every connection, with the required isolation and busy behavior.
3. Trace read-modify-write and claim paths. Use atomic predicates/transactions and expiry/fencing semantics for concurrent workers; do not hold write locks across network or user think-time.
4. Use parameterized values and explicit identifier allowlists. Inspect query plans and index/FTS/JSON maintenance on representative data.
5. Verify affected contention, migration rollback, crash durability, backup and restore behavior. Use file-backed tests for promises involving locking or WAL.

Scope and completion

Deliver the affected configuration/transaction contract and observed checks. A query repair does not require database redesign. Apply migration procedures only when stored state must survive; follow the project retention/reset policy. Encryption and retention promises need independent deployment checks.

Use the user’s existing intent and authorization. Ask only for missing information that materially changes the result; use additional reviewers when they address a concrete uncertainty. Treat unavailable checks as gaps rather than successful verification.

Focused references