eclaire-labs/eclaire v0.6.0

on GitHub

Highlights

• Unified deployment: frontend + backend + workers can run in a single container
• Simplified Self-Hosting - new one-command setup.sh flow, plus a streamlined compose.yaml
• Better AI Support - New vision models, llama.cpp router, expanded provider support
• Modern Frontend - Migrated from Next.js to Vite + TanStack Router
• New Admin CLI - Manage your instance from the command line

Features

• Unified Deployment: Single container can serve as backend, workers, or both via SERVICE_ROLE environment variable
• SQLite Support: Full SQLite database support alongside Postgres with comprehensive parity tests
• Database Queue Mode: Use Postgres or SQLite for job processing instead of Redis/BullMQ
• In-Memory Notifications: Single-process deployments no longer require Redis
• Admin CLI: New admin-cli integrated into Docker for instance management
• Auto-Upgrade System: Database migrations run automatically at startup
• Qwen3-VL-8B: Added support for Qwen3-VL-8B vision model
• llama.cpp Router: Support for llama-server's new router endpoint
• Request ID Tracing: Better observability and debugging across distributed components
• Version-Prefixed Encryption: Enables future key rotation support

Improvements

• Frontend: Migrated from Next.js to Vite with TanStack Router for faster builds and modern routing
• ES Modules: Complete migration from CommonJS to ES modules
• Tailwind CSS v4: Upgraded to latest Tailwind with tw-animate-css support
• Transaction Support: Read-Modify-Write for Postgres, mutex serialization for SQLite
• Services Architecture: Thin routes pattern with extracted services layer
• Modular Packages: New @eclaire/ai, @eclaire/storage, @eclaire/queue packages
• AI Tool Calling: More robust native tool calling support
• AI Providers: Improved support for llama.cpp, MLX-LM, MLX-VLM, and LM Studio backends
• Dependency Upgrades: Vite 7, Vitest 4, Playwright 1.57, Pino 10, Recharts 3, Better Auth 1.4.5

Bug Fixes

• User queue jobs now deleted when account is deleted
• Fixed AI response truncation/repetition detection with improved JSON parsing
• SQLite case-insensitive sorting for text columns
• Docker layer caching optimized for faster builds
• Queue callback race conditions resolved with timeout safety guards
• Route detail views render correctly
• Image URL fallback uses thumbnailUrl consistently
• Auth session tests include Origin header for CSRF validation
• Sharp/native dependency handling with SHARP_IGNORE_GLOBAL_LIBVIPS

Breaking Changes

• TaskStatus: Removed cancelled status from enum
• SERVICE_ROLE: New environment variable for deployment mode configuration
• Data Directories: Changed from data/db to data/postgres, data/pglite, data/sqlite
• SQLITE_DB_PATH: Renamed from SQLITE_DATA_DIR
• Environment: Simplified to single .env file configuration
• Admin CLI: model-cli replaced with admin-cli
• Build Script: build.sh now defaults to dev mode (use --prod for production)
• Production Port: Standardized to port 3000

Migration from v0.5.x

There is no automated upgrade path from v0.5.x to v0.6.0. Due to significant architectural changes, we recommend setting up a fresh v0.6.0 instance and transferring your data from your previous installation.

If you need help migrating, please open an issue or reach out to us - we're happy to assist.

Infrastructure

• All dependencies pinned to exact versions for reproducible builds
• Root biome config for consistent formatting/linting
• Restructured tests directory (__tests__ → tests)
• TypeScript config standardized (typecheck includes tests, build excludes)
• ioredis pinned to 5.8.2 for BullMQ compatibility

Documentation

• AI Model Configuration guide with Qwen3-VL-8B examples
• First login instructions added to README
• Clarified WORKER_AI_LOCAL_PROVIDER_URL documentation