Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.4.0] - 2026-08-24

DuckDB Vectorized Analytics Engine: Integrated DuckDB in StatisticalAnalyzer (src/core/domain_services.py) for high-performance vectorized descriptive statistics with automatic fallback to Pandas/NumPy.
Async Background Job Queue System: Added JobRepository (src/adapters/repositories/job_repository.py) and endpoints POST /api/jobs/execute (returning 202 Accepted + job_id) and GET /api/jobs/{job_id} for polling background execution.
CSV Formula Injection Sanitization: Implemented cell-level sanitization in FileSystemAdapter.export_file to escape =, +, -, @ formula triggers in exported CSV and Excel files.

Frontend Plotly Code-Splitting: Created LazyPlot.tsx using dynamic Promise.all([import(...)]) for Plotly.js, reducing the initial JavaScript bundle from 4.8MB to 239KB.
Virtualized Data Table: Created VirtualizedTable.tsx for paginated (10, 25, 50, 100 rows) and scroll-limited rendering of tabular previews.
Health Probes: Added /healthz (liveness) and /readyz (readiness) endpoints in src/adapters/api/router.py.

Hexagonal Architecture Core Ports: Added FileIOProvider and PlotterProvider abstract ports to src/core/ports.py to enforce strict hexagonal layer isolation.
Outlier Handling Skill: Registered @register_skill("handle_outliers") in clean_skills.py for IQR detection, removal, and Winsorization capping.
Extended Test Suite: Added 32 new tests across tests/test_skills.py, tests/test_domain_services_edge_cases.py, and tests/test_api_edge_cases.py, increasing test count from 21 to 53 passing tests and test coverage to 88%.
GitHub Actions CI Pipeline: Created .github/workflows/ci.yml running Python 3.11 linting (Ruff), type checking (Mypy), test coverage (Pytest), and Node.js 20 frontend builds.
Global Pytest Initialization: Added tests/conftest.py and src/core/agents/skills/__init__.py for automatic skill registration upon package import.

Strict Layer Isolation: Removed all infrastructure imports (src.adapters.*) from core skills (io_skills.py, visualization_skills.py), replacing them with dynamic dependency resolution via core provider ports.
Router Consolidation: Refactored /api/outliers and /api/plot endpoints in src/adapters/api/router.py to delegate execution through agent_manager.execute_skill(...).
Payload Extraction Refactoring: Added _prepare_session_dataframe helper in src/adapters/api/router.py to remove duplicate session extraction logic across endpoints.
Statsmodels Fallback: Added robust linear regression fallback in PlottingAdapter.plot_regression when statsmodels is omitted or uninstalled.
Code Quality & Type Annotations: Fixed all Ruff linting warnings and updated Mypy configuration in pyproject.toml with clean strict compliance.

Fixed Vercel stateless execution payload extraction for /api/clean/nulls and /api/clean/scale.
Fixed Vercel /tmp export path handling in FileSystemAdapter.export_file.

Initial Vercel serverless deployment setup with FastAPI backend and React frontend.
Native NumPy implementations for K-Means clustering, Z-Score scaling, and Kurtosis distribution shape calculations.
AgentManager orchestration and 9 super-skills system.