Skip to content

CHANGELOG.md

Latest commit

 

History

History
89 lines (74 loc) · 6.05 KB

CHANGELOG.md

File metadata and controls

89 lines (74 loc) · 6.05 KB

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.

Unreleased

Added

  • AI Code Review: Support for OpenAI, Anthropic (Claude), Ollama, Google Gemini, and Azure OpenAI

  • File Context: Fetch full file content for better AI review context

  • Chunked Review: Automatically split large MRs/PRs into batches

  • Smart Filtering: Auto-skip config files, lock files, and generated files

  • Multi-Platform Webhooks: Unified webhook for GitHub, GitLab, and Bitbucket

  • Real-time Updates: SSE-powered live status updates

  • Dashboard: Visual statistics and metrics

  • Review History: Track all code reviews with detailed logs

  • Project Management: Manage multiple repositories

  • Member Analysis: GitHub-style contribution heatmap and statistics

  • LLM Configuration: Configure multiple AI models

  • Prompt Templates: System and custom prompt templates

  • IM Notifications: DingTalk, Feishu, WeCom, Slack, Discord, MS Teams, Telegram

  • Daily Reports: Automated daily code review summary with AI analysis

  • Git Credentials: Auto-create projects from webhooks

  • System Logging: Comprehensive logging for all events

  • Authentication: Local and LDAP authentication

  • Role-based Access Control: Admin and User roles

  • Multi-Database Support: SQLite, MySQL, PostgreSQL

  • Async Task Queue: Optional Redis-based async processing with SSE notifications

  • Sync Review API: Synchronous review endpoint for Git pre-receive hooks

  • Internationalization: English and Chinese support

  • Responsive Design: Mobile-friendly interface

  • Dark Mode: Light and dark theme toggle

  • Global Search: Cross-project search for reviews and projects from the header

  • Multi-Language Review Prompts: Auto-detect programming language from diffs and inject language-specific review guidelines (12+ languages)

  • Batch Operations: Batch retry and batch delete for review logs

  • SSE Notification Bell: Real-time notification bell with unread badge and live review events

  • Reports API: Weekly/monthly report with period comparison, daily trends, and author rankings

  • Issue Tracker Integration: Auto-create Jira, Linear, GitHub Issues, or GitLab Issues when review score is below threshold; now wired into review flow with per-tracker threshold and test connection support

  • Auto-Fix PR: AI-powered code fix generation — automatically creates a fix branch, commits AI-generated patches, and opens a PR (GitHub) or MR (GitLab) for low-score reviews

  • Rule Engine: Automated CI/CD policies with conditions (score_below, files_changed_above, has_keyword) and actions (block, warn, notify)

  • Prometheus Metrics: /metrics endpoint for monitoring

  • Audit Logging: Automatic audit logging for all admin write operations

  • Review Diff Cache: SHA-256 hash deduplication to skip already-reviewed diffs

  • CSV Export: Export review logs as CSV for offline analysis

  • RBAC Enhancement: Three roles (admin, developer, user) with granular permissions

  • Project Members: Per-project user binding with owner/maintainer/viewer roles

  • SSE Heartbeat: 30-second heartbeat to prevent Nginx 504 timeouts on SSE connections

  • Manual Score Override: Admin can manually override AI review scores with reason tracking and original score preservation

  • Auth Refresh: Refresh-token based session renewal (POST /api/auth/refresh) with httpOnly cookie storage

  • Refresh Token Rotation: Server-side refresh token storage with rotation and revocation support

  • Proactive Session Refresh: Frontend proactively refreshes access token before expiration to reduce 401 interruptions

  • Auth Session Settings UI: Admin settings page now supports configuring access/refresh token expiration

Changed

  • Unified API Response: All handlers migrated to pkg/response package with standardized {code, data, message} envelope
  • Structured Logging: All 16 service files migrated from log.Printf to pkg/logger (zerolog-based logger.Infof/Errorf/Warnf)
  • Models Split: Decomposed monolithic models.go (290 lines) into 13 individual model files for better maintainability
  • Main Entry Split: Split main.go into bootstrap.go (initialization) and routes.go (routing)
  • Frontend API Layer: Added automatic response envelope unwrapping in axios interceptor
  • Notification Refactor: Refactored notification service using Strategy Pattern with NotificationAdapter interface, reducing ~600 lines to ~90 lines + 8 platform adapters
  • DB Query Optimization: Consolidated 9 separate DB queries in daily report stats collection into a single aggregate SQL query
  • Regex Pre-compilation: Moved regex patterns in ai.go and file_context.go to package-level pre-compiled variables
  • Function Extraction Dedup: Extracted shared matchBoundariesToRanges helper to eliminate ~80 lines of duplicated code across 4 language extractors
  • HTTP Timeout: Added 10-second timeout to notification HTTP client to prevent goroutine leaks
  • Dockerfile Hardening: Upgraded to Node 22 & Alpine 3.21, added non-root user, added -ldflags="-s -w" for ~30% smaller binary
  • Vite Code Splitting: Added manualChunks configuration to split vendor dependencies (antd, react, recharts, utils) into separate cacheable bundles
  • Dependency Upgrades: Updated anthropic-sdk-go (v1.19→v1.26), genai (v1.43→v1.47), ollama (v0.15→v0.17), asynq (v0.25→v0.26)

Fixed

  • Feedback Score Parsing: Allow 0/100 as a valid score in AI feedback responses
  • Unreviewable Status: Set review_status to unreviewable when feedback score is 0 (e.g., diff unavailable)

Security

  • JWT-based authentication
  • Refresh-token based session renewal with httpOnly cookie
  • LDAP integration support
  • Webhook secret validation
  • Duplicate review prevention for same commit across branches