featcat v0.2.4

Fixed

• CLI remote mode: ask, discover, doc generate, monitor check, and monitor baseline now call server API endpoints when FEATCAT_SERVER_URL is set, instead of trying to connect to local Ollama
• RemoteBackend: Added ai_ask, ai_discover, doc_generate, monitor_check, monitor_baseline methods for server-side operations

featcat v0.2.3

Fixed

• Discovery plugin: Handle LLM returning feature names as strings instead of dicts
• CLI remote mode: doctor and job commands now work when FEATCAT_SERVER_URL is set
• Publish workflow: Fix TestPyPI verification with uv index-strategy

featcat v0.2.2

Fixed

• Discovery plugin: Handle LLM returning feature names as strings instead of dicts
• Publish workflow: Pin version when installing from TestPyPI to avoid stale cache

featcat v0.2.1

Fixed

• CLI remote mode: doctor command now checks server connectivity when FEATCAT_SERVER_URL is set instead of only checking local SQLite
• Job CLI remote mode: All featcat job commands (list, logs, run, enable, disable, schedule) now call the server API in remote mode instead of requiring local SQLite access

featcat v0.1.0

Initial Release

Added

• Catalog: Register data sources (local + S3/MinIO), scan Parquet files, auto-extract schema and statistics
• AI Discovery: Describe a use case, get ranked feature recommendations and new feature suggestions via LLM
• Auto-doc: LLM-powered documentation generation for features, with batch processing and Markdown export
• NL Query: Natural language search across the feature catalog, with Vietnamese language detection and fuzzy fallback
• Quality Monitoring: PSI drift detection, null spike alerts, range violation checks, zero variance detection
• LLM Backends: Ollama and llama.cpp server support with streaming, retry logic, and JSON extraction
• Response Caching: SQLite-backed LLM response cache with configurable TTL per plugin
• TUI: Terminal UI with dashboard, feature browser, monitoring view, and AI chat (Textual)
• S3 Support: Read Parquet metadata directly from AWS S3 and MinIO via PyArrow S3FileSystem
• CatalogBackend Interface: Abstract backend supporting local (SQLite) and remote (HTTP) modes
• CLI: Full command suite — init, source, feature, discover, ask, doc, monitor, doctor, stats, export, cache, ui
• CI/CD: GitHub Actions with lint, type-check, test (Python 3.10–3.13), build verification, and PyPI publishing
• Documentation: English (default) + Vietnamese, with setup guide, user guide, and admin guide