What's Changed
Fixes
- Fixed SQL Copilot retry logic to prevent infinite retries when generating read-only SQL statements.
- Resolved styling inconsistencies in Copilot-generated SQL result components, including tooltip alignment and code block rendering.
- Fixed AI action failures caused by missing column metadata in schema context prompts.
- Corrected database summary page layout issues, improving visual alignment of metadata cards and tooltips.
- Fixed connection establishment errors in end-to-end tests and the connection flow UI.
- Addressed app sidebar icon misalignment and inconsistent update button styling across navigation sections.
Testing & CI Improvements
- Added comprehensive end-to-end tests covering SQL Console, Connections, and Auth flows.
- Introduced a new GitHub Actions workflow for automated E2E test execution.
- Fixed multiple Playwright configuration issues, including
.tmpdirectory filtering and authentication state handling. - Stabilized flaky E2E test suites (e.g., login, workbench, connection setup) through targeted fixes in test helpers and auth setup.
Documentation & Tooling
- Updated Docker run command in
README.mdto remove obsolete environment variables (BETTER_AUTH_URL,DB_TYPE). - Added an AI-powered release notes generator with dedicated GitHub Action and CLI script.
Internal
- Refactored and stabilized Playwright test infrastructure, including auth persistence, config, and helper utilities.