Changelog
Track the journey of the next-generation database management IDE.
v1.6.7
Production-grade CockroachDB & MariaDB, Adversarial Testing & a Hardened Web Platform
CockroachDB and MariaDB promoted to production-ready
CockroachDB now gets a dedicated Postgres-derived strategy with the 26257 default port and crdb_internal-based metrics instead of PostgreSQL-only SQL; MariaDB gains views, functions, function parameters, and a guard that blocks empty updates before invalid SQL is generated.
Shared TLS contract for MySQL/MariaDB and Redis
The connection form now has a TLS toggle for MySQL and MariaDB to disable SSL for servers that reject handshakes; Redis connections persist a proper TLS flag too, with a backfill migration preserving existing auto-TLS behavior.
Adversarial pentest suites for SQL/NoSQL guards and the HTTP layer
New attack suites simulate bypasses via identifier smuggling, IPv4-mapped IPv6, and CSRF header spoofing; validateHost is now fail-closed for empty hosts, with property-based fuzzing covering all inputs.
Hardened web platform: PWA, sitemap, security.txt, and billing e2e
Added a PWA manifest, a sitemap with dynamic lastmod, a security.txt for responsible disclosure, an html lang synced to the app language, and an e2e flow for the MoMo/ZaloPay billing return page.
Indexed NoSQL sidebar search and query duration display
The NoSQL sidebar navigates through a new search index, the search service was refined, and query results now show execution duration (durationMs) for direct performance visibility.
v1.6.6
Observability, Query Control & Supply-chain CI
Pino structured logging with per-request correlation IDs
Production runs pino structured logging with correlation IDs on every request for end-to-end tracing; development keeps the familiar Nest default logger.
Running-query cancellation & the Active Queries panel
Each engine has its own cancellation path (pg_cancel_backend, KILL QUERY) so cancels happen on the server; the Active Queries panel lists running queries and cancels them individually.
SQL-keyed AI Explain cache & follow-up prompts
AI query explanations are cached by SQL content so re-asking the same query costs nothing extra; each explanation supports inline follow-up questions.
Supply-chain CI: Trivy, SBOM, pinned images & Docker CVE patches
The containers pipeline adds Trivy vulnerability scanning, SBOM generation, and pinned base images; the runtime image strips the npm toolchain and upgrades OS packages to patch CVE-2026-73566.
Nightly benchmarks, Firefox E2E & the IPv4-mapped IPv6 SSRF fix
Added nightly benchmarks, an expanded Playwright matrix including Firefox, and property-based guard fuzzing; also discovered and fixed an SSRF bypass via IPv4-mapped IPv6 addresses.
Localized server security messages & React correctness fixes
Server-side security messages are localized or normalized, hardcoded Vietnamese strings in the AI chat UI moved into i18n, alongside QueryResult typing and React purity-violation fixes.
v1.6.5
Hardened AI Providers, Race-safe Billing & Production-safe DataGrid
Custom AI providers persist, verify, and get a redesigned form
User-added OpenAI-compatible providers now persist reliably, verify against the endpoint before use, and the configuration screen was redesigned with stricter per-field validation.
Native PDF delivery & Google GenAI SDK migration
Attached PDF documents are passed through natively to capable providers instead of text extraction, and the entire Gemini lane migrated to the new Google GenAI SDK for driver-version alignment.
Surfaced model capabilities & preflight checks
The model catalog now reflects real capabilities (vision, PDF, live search) matching the configured provider, availability is preflighted before requests, and AI explanations offer a safe query-correction handoff instead of blind execution.
Hardened auth & atomic multi-statement transactions
Token refresh and OAuth identity linking were tightened against race conditions, billing activation became race-safe, and multi-statement SQL scripts now run inside a single transaction — a mid-script failure rolls everything back.
Production-safe CSV DataGrid & recoverable error boundary
CSV import/export now handles encoding, delimiters, and large payloads correctly; the app gained a recoverable error boundary instead of a white screen; dashboards preserve the previous snapshot when a refresh fails.
v1.6.4
Flexible AI Configuration, Smarter Query Guards & Broader Localization
Role-based AI configuration & custom providers
Added a dedicated AI configuration hub for Assistant, Explain, SQL, and NoSQL model selection, plus custom OpenAI-compatible providers with their own default models.
Higher-signal query safety warnings
Tightened query analysis so warnings focus on genuinely risky actions, cut down false alarms on routine creation flows, and explain the reason more clearly before execution.
Smoother AI Explain and SQL Steps inside the editor
Added AI-powered query explanation, an SQL Steps dialog for reordering and running multi-step SQL sequences, and toolbar refinements that make the editor workflow feel more cohesive.
Hot-path performance backed by real benchmarks
Accelerated table-window browsing, adaptive migration batching, search indexing, and other critical data-access paths while publishing a real performance baseline for future tuning passes.
Deeper bilingual coverage for workspace, docs, and landing
Expanded the shared content map and i18n coverage so more workspace, documentation, and landing surfaces stay semantically consistent in Vietnamese and English.
v1.6.3
Unified AI Outputs, Actionable NoSQL & Leaner Runtime
Shared structured outputs across chat, SQL, and NoSQL
Unified structured outputs so freeform chat, SQL generation, and NoSQL assistance behave more consistently while reducing one-off parsing paths and making results easier to reuse.
NoSQL summaries, mutations, and deeper analysis
Expanded the NoSQL workspace with inline summaries, clearer mutation support, and reworked schema and visualize surfaces that make data behavior easier to inspect.
Leaner provider routing with broader extensibility
Simplified the provider runner, consolidated OpenAI-style compatibility logic, and widened model lanes so adding providers or swapping models depends less on special-case branches.
More stable conversation replay and prompt context
Strengthened conversation replay, context assembly, and prompt building so the Assistant keeps its flow more reliably during longer exchanges or when switching between SQL and NoSQL.
Major refactors across NoSQL and visualize screens
Reworked the aggregation, schema analysis, and visualize screens to reduce scattered logic, improve separation of concerns, and prepare the ground for later UX iterations.
v1.6.2
Expanded AI Vision & New Model Catalog
Image attachments & vision-aware routing
AI chat now accepts direct image attachments in prompts and automatically prioritizes vision-capable lanes for screenshots, diagrams, and image-based tasks.
Beeknoee added to routing
Added Beeknoee as a first-class routing provider with dedicated environment configuration instead of hiding it behind implicit fallback paths.
Dedicated model catalog & MiniMax M2.7
Split the AI Assistant model list into a dedicated catalog and expanded the picker with Qwen 3 235B, GLM 4.7 Flash, and MiniMax M2.7.
Polished AI chat composer
Refreshed the AI chat composer with better styling, smoother animation, clearer model icons, and a more expressive prompt composition experience.
60-second AI timeouts for heavier payloads
Raised provider timeouts to 60 seconds so larger prompts, heavier schema context, and image payloads complete more reliably.
v1.6.1
Real Billing, Tighter Access Control & Richer AI Context
Real billing checkout with MoMo/ZaloPay
Replaced mock pricing with a real checkout flow including plan definitions, subscription persistence, payment webhooks, and billing return handling.
Resource-level permissions & security hardening
Tightened access to connections, queries, and dashboards using resource-level policies while further hardening SSH tunnel SSRF protection and sensitive AI paths.
Faster, deeper DataGrid workflows
Added a cell inspector, bulk replace, search highlighting, smoother scrolling, stronger NoSQL virtualization, and smarter metadata/query caching.
Smarter schema-aware AI context
AI context now includes sample rows, NoSQL-specific rules, overflow guards, and more accurate autocomplete thanks to stronger type awareness.
End-to-end vi/en localization
Rolled out i18n across the client, server, email templates, and auth flows so key UI, errors, and notifications can stay consistent in Vietnamese or English.
v1.6.0
AI Foundational Architecture & Reasoning Transparency
AI Reasoning Transparency
Added a collapsible Thought block in the chat interface to show the AI's summary and reasoning process, improving transparency and trust.
Dynamic AI Routing & Fallback
Refactored the backend with routing and provider runner modules to support diverse model networks like OpenRouter and Groq, plus smoother fallback behavior when the primary model is overloaded.
NoSQL AI Assistant Integration
Introduced the new NoSqlAiQueryBox, bringing completion, syntax repair, and query generation directly into the MongoDB workspace.
Chat State & UI Overhaul
Overhauled the conversation state flow using aiChatSlice, useAiChat, and a new AiChatInput component to support longer-lived context and smoother multi-line input.
Autocomplete Hardening
Tightened prompt context, removed UI hang scenarios with stricter timeouts, and aggressively prioritized ultra-fast models like Gemini Flash Preview.
v1.5.0
NoSQL Explorer Parity & UX Refinement
NoSQL Workspace Parity
Brought shortcuts, File/Edit menus, and Visualize/Diagram tools to NoSQL for a more consistent workspace experience with SQL.
Version History & Metadata Freshness
Added entity version history and automatic metadata freshness tracking so data stays accurate.
Org Backup & Migration Preview
Introduced organization backups and migration preview tooling to make infrastructure transitions safer.
Real-time Presence Tracking
Shows who is active in the workspace in real time through presence tracking.
Persistent Workspace State
Persists result panel height, Tree/Grid view mode, and aggregation stages across page refreshes.
Universal Visualize Hub
The visualization page now fully supports NoSQL, enabling smoother chart creation from MongoDB data.
v1.4.0
Advanced Security & Infrastructure Hardening
Enterprise-Grade Security Hardening
Strengthened SQL Guard against EXEC bypasses, blocked SSH tunnel SSRF vectors, and tightened team privacy logic.
Advanced Data Privacy
Sanitized detailed database errors to prevent infrastructure leakage and further secured default admin credentials.
Migration Guardrails
Applied safety timeouts to large migration flows to prevent resource exhaustion.
Pentest & Security Validation
Completed an advanced security audit with real-world logic attack scenarios.
v1.3.1
Team Collaboration & Mobile Polish
Organizations & Shared Workspaces
Added the collaboration foundation for organizations, members, roles, and shared connections, queries, and dashboards.
Mobile Collaboration Access
Teams are now reachable from the avatar menu on mobile, while workspace, dialog, and admin screens are tuned for smaller layouts.
Connection Diagnostics
ConnectionDialog now includes Test Connection alongside UI fixes and a smoother connection setup flow.
Realtime Stability
SSE notifications, cache, and background update flows were realigned with the current backend API and Redis infrastructure.
v1.3.0
Next-Gen AI & Comprehensive Type Safety
AI Service Decomposition
Decomposed the AI service into specialized sub-services for routing, provider runners, chat completion, and autocomplete.
Complete Type Safety
Removed any types, constrained generics, extracted AI constants, and improved code quality using SOLID, KISS, YAGNI, and DRY.
Redis Caching & Global Search
Added query result caching, global schema search, real-time SSE notifications, and more efficient session management.
AI-Powered SQL Generation
Added SQL generation from natural language plus semantic search for database objects.
v1.2.0
Smart ERD & Real-time Notifications
ERD Performance Optimization
Improved ERD performance and visual clarity with Redis caching and smart suggestions.
Real-time SSE Notifications
Added real-time notifications via Server-Sent Events for query status and system updates.
Client-side Testing Suite
Added unit tests for ConnectionService and SavedQueryService to improve reliability.
v1.1.0
UI Upgrade & Hooks Refinement
Cinematic Dark Mode Landing Page
Upgraded the landing page with a cinematic dark theme, smoother animation, and richer visual effects.
Query ERD Hooks Alignment
Aligned Query and ERD hooks, fixed TypeScript errors, and cleaned up unused code.
v1.0.5
Code Quality & Type Fixes
SOLID, KISS, YAGNI, DRY Refactoring
Improved code quality with software design principles focused on clarity, boundaries, and maintainability.
TypeScript Error Resolution
Fixed TypeScript errors, cleaned up unused props, and tightened types.
v1.0.0
The Core Platform Genesis
NestJS & Prisma Backbone
Shipped a stable backend supporting OAuth, JWT, and AES-256-GCM encryption.
Auth Service Specification
Added the auth service specification and agent directory configuration.