Logs perubahan

Kemas kini dan penambahbaikan terkini

2.0.0

Agent-Native Intelligence, OAuth Apps & CDN Reliability

MCP — Smarter AI Agent Tooling

Better i18n's MCP server now helps AI agents avoid common mistakes and clean up data issues autonomously.

Semantic Duplicate Detection `NEW`

When an AI agent creates a key with `createKeys`, the response now warns if the same source text already exists in another namespace. For example, creating `actions.removeItem` with value "Sil" when `common.delete` already has the same text will return a reuse hint. This prevents translation fragmentation where the same phrase gets translated independently across namespaces — saving AI costs and keeping translations consistent.

Pass `force: true` to skip the check for fast bulk imports.

Phantom Key Detection `NEW`

`listKeys` now flags legacy duplicate rows (`p: true`) that shadow proper namespaced keys and corrupt CDN files during publish. AI agents can call `deleteKeys` with the flagged IDs, then `publishTranslations` to fix CDN corruption — no database access needed.

`createKeys` also warns when a phantom exists for a key being created, with actionable cleanup instructions in the `warnings[]` response.

Translation Context v2 — Per-Key RAG `NEW`

`getTranslationContext` now accepts a `keys` parameter with up to 50 key UUIDs. The response includes pgvector-retrieved passages (glossary terms, past translations, project instructions) ranked by cosine similarity per key — giving AI agents the same context the platform's AI drawer uses.

Other MCP Improvements

---

  • `setTranslations` — New bulk write tool with ~55-65% smaller payload than `updateKeys` for multi-language batches
  • `cancelSync` — Abort pending publish jobs before they reach CDN
  • `getSync` with `waitMs` — Server-side blocking wait replaces 3-6 polling round-trips
  • CDN format metadata — `fileStructure` and `keyFormat` fields in `getProject` and publish hints

OAuth App Installation System `BETA`

A GitHub-Apps-style OAuth 2.0 flow for partners to integrate Better i18n into their platforms.

---

  • Scope-based authorization with Cloudflare-style two-step consent UI
  • Installation tokens — short-lived, project-scoped API access
  • Connected Apps management in user settings with bulk revoke
  • Security email notifications on grant create and revoke
  • Partner API — project creation, language management, and translation access

CDN Publish Pipeline Hardening

April saw significant reliability improvements to the sync worker — the most critical service in the platform.

Hash-Based CDN Diff `NEW`

Publish now computes SHA-256 hashes of generated files and only uploads files whose content actually changed. For the landing project (2,310 CDN files), this dropped per-publish R2 writes from 2,310 to ~22 — a 99% reduction in CDN write operations.

Source Author Attribution `NEW`

Translation keys now carry their Git author (name, email, avatar) from the commit that introduced them. Visible in the dashboard with avatar display.

Namespace CDN Delivery `NEW`

Projects can now serve translations as one file per namespace (`/en/auth.json`, `/en/common.json`) instead of a single monolithic file. Configurable via `fileStructure` (single_file / namespaced_folders) and `keyFormat` (flat / nested).

Other Pipeline Fixes

---

  • Phantom row skip — File generator now deterministically skips legacy duplicate rows that caused random CDN value corruption
  • Slim manifest — Per-locale namespace objects removed from manifest, reducing payload size
  • Queue concurrency — `max_concurrency` set to 5 to eliminate head-of-line blocking
  • Parameter overflow protection — Large ID arrays chunked to stay under PostgreSQL's 65K limit
  • Status update batching — Batch size increased 100→2,000 for large projects

Platform Improvements

  • Unified quota enforcement — Vercel-style billing UI with plan limits enforced across all surfaces (MCP, dashboard, API)
  • Context-aware AI suggestions — AI drawer now shows contextual suggestions based on the current page
  • Content CMS webhooks — Comprehensive event dispatch for translations, keys, languages, and sync operations
  • Path collision detection — `createKeys` now detects leaf↔object conflicts in nested JSON before they reach CDN
  • Admin dashboard — Real CDN analytics, queue monitoring, and rate limiter panels wired to Cloudflare APIs
1.6.0

Bulk Operations, Navigation Redesign & Content API Keys

Translation Editor

  • Bulk translation workflow: Select, approve, reject, or translate multiple translations at once with Shift+click range select and batch operations.

Content CMS

  • Relation fields & built-in Users model: Content models now support relation fields with proper editor rendering and built-in author references.
  • Content API keys with integration guide: Quick Start section with cURL examples, SDK install commands, and code snippets after creating an API key.
  • Cloudflare AI migration & Content MCP: Faster AI translations powered by Cloudflare AI with model-based authorization and llm.txt endpoints.

Dashboard

  • Interactive AI chat in Quick Start: Try the AI assistant directly from the onboarding flow without leaving the setup wizard.
  • Project navigation redesign: Cleaner tab structure with Languages merged into Overview and settings reorganized with sidebar layout.
  • Redesigned "Add Languages" flow: Popular languages section, search-as-you-type filtering, and regional variant grouping.

Bug Fixes & Improvements

  • FIXEDTeam Members permissions: Admins can no longer assign Owner role or remove other admins. Pending invites count now displays correctly.
  • FIXEDUsage Overview: Translation Keys and AI Usage values now pull real data from the API.
  • FIXEDContent editor crash: Fixed "Expected drag drop context" error on entry detail pages.
1.5.0

Content Management, AI Translations & Onboarding

Content CMS

  • Blog, changelog and roadmap content: Create and edit content entries with full localization support.
  • Rich text editor: Writing experience powered by PlateJS with formatting, code blocks, and more.
  • Multi-language content translations: Manage translations for each content entry across all project languages.
  • Draft/published status control: Control when content goes live with status management.
  • Unsplash integration: Search and add free stock images directly from the editor.
  • Auto-save: Never lose your work with automatic saving.

AI Translations

  • Auto-approve mode: Complete the entire translation process with a single click.
  • Faster translation suggestions: Improved AI models deliver better quality translations.

Dashboard

  • Help menu: Quick access menu for support and resources.
  • Getting started guide: Step-by-step setup wizard with estimated time and notes for each step.
  • Visual progress indicators: Track completed onboarding steps at a glance.
  • Drag-and-drop language ordering: Arrange languages in your preferred display order.
  • Reset language order: Quickly reset to default ordering with English on top.
1.4.0

AI Efficiency, Tool Decomposition & CDN Architecture

AI Agent

  • Optimized AI responses: All read endpoints now return 40–50% smaller responses, reducing AI token costs significantly.
  • Optimized tool output: `getKeyDetails` returns a more efficient format (55% token reduction) with built-in field legend.
  • Immediate HTTP streaming: AI chat opens the HTTP connection instantly, eliminating perceived "hanging" during initialization.
  • AI context window: Token counting now uses last step usage instead of cumulative totals.
  • Auto-approve UX: Clearer auto-approve prompt flow with semantic XML tags.

Dashboard

  • Note activity type: Annotate translation keys with notes tracked in the activity feed.
  • Billing cycle toggle: Monthly/yearly pricing with 20% yearly savings badge.
  • Tool UI decomposition: Monolithic tool.tsx decomposed into 11 focused files with zero visual changes.

Infrastructure

  • R2 dual-bucket architecture: Flag images and static assets migrated to dedicated R2 bucket (`s3.better-i18n.com`), separate from translation files.
  • Publish query optimization: Parallelized queries for faster response times.
  • User attribution: Sync jobs now track which user triggered them.

Bug Fixes & Improvements

  • FIXEDMCP publish: Fixed failing when CDN is enabled but no files uploaded yet.
  • FIXEDSoft-delete rename: Fixed transaction using single connection for find+rename+insert.
  • FIXEDVList height: Fixed height mismatch for small translation lists (<20 items).
  • FIXEDCache invalidation: Fixed organization settings cache invalidation key.
1.3.0

MCP Publish Workflow & Reliability Fixes

MCP Tools

  • MCP publish tools: Two new tools — `getPendingChanges` to preview deployments and `publish` to deploy translations to CDN, all from your IDE.
  • Complete MCP workflow: AI assistants can now follow the full cycle: read → translate → verify → publish → confirm.
  • Pending publish hints: Write operations now return hints reminding AI to call publish when changes are ready.
  • MCP router documentation: Comprehensive tool documentation with categories and recommended workflow guide.

CDN & Publishing

  • Default CDN format: Changed default from JSON_FLAT to JSON_NESTED for better i18n library compatibility.
  • CDN fallback by default: New projects automatically get fallback language support enabled.

Dashboard

  • Safer key recreation: Fixed soft-delete rename with single database transaction, preventing conflicts.
  • Better chat compaction: `/compact` command now shows user message and actual summarization content.
  • Language sorting: Languages are now sorted by status: active → draft → archived.

Bug Fixes & Improvements

  • FIXEDSlash command auto-submit: Fixed double submission issue.
  • FIXEDSummarization cleanup: Removed deprecated summarization tool from server-side.
1.2.0

Namespace Context, Fallback Languages & Open Source

AI & MCP

  • Namespace context for AI: AI models can now attach rich context (description, team, domain, AI prompt, tags) when creating or updating keys.
  • Smarter key mentions: Keys mentioned in AI chat now include their namespace context for precise AI targeting.
  • Faster Gemini streaming: Optimized system prompts for significantly faster streaming with Gemini models.

SDKs & Platform

  • Localization fallback: When adding a new language, missing keys automatically inherit from the fallback language.
  • Open source release: Packages, MCP server, and landing page now public at github.com/better-i18n/oss.
  • Relational namespaces: Translation keys now have proper relational connection to namespaces for better organization.

Dashboard

  • Redesigned project settings: Completely revamped project settings interface for better usability.
  • Enhanced filter UI: Improved filtering experience across the translation dashboard.
  • CDN republish support: Deleting CDN files now resets published status, allowing clean republishing.
1.1.0

AI Performance Boost & Smoother Streaming

AI Agent

  • Smoother AI responses: Upgraded AI SDK to v6.0.39 with smoothStream for fluid, natural response streaming.
  • Smart context caching: 30-second cache for project context queries reduces database load and speeds up AI responses.
  • Smarter AI thinking: Upgraded thinking configuration for more accurate translation decisions.

MCP & Developer Tools

  • Better MCP tool schemas: Redesigned MCP tool input format for more intuitive IDE integration.
  • Instant MCP translations: Keys created via MCP are now automatically approved.
  • Enhanced key mentions: Namespace field added to key mentions in AI chat for better context awareness.

Dashboard

  • Saved view descriptions: Add descriptions and privacy settings to your saved translation views.
  • Cleaner chat UI: Simplified copy button design for less visual clutter.

Bug Fixes & Improvements

  • FIXEDMessage parsing: Fixed message parsing removing debug console.logs.
  • FIXEDPublish reliability: Improved publish batch processor reliability.
  • FIXEDData integrity: Enhanced soft-delete helper functions for better data integrity.
1.0.0

Smarter AI Agent & MCP Tools

AI Agent

  • MCP delete keys tool: Soft delete translation keys directly from your IDE using the MCP server.
  • AI agent performance: Faster response times and better context handling.
  • Chat summarization: Improved conversation summarization for long sessions.
  • Gemini integration: Migrated to new thinkingBudget API for better AI responses.

Dashboard

  • Developer tools modal: New modern DevTools with chat sessions viewer and network inspector.
  • Admin impersonation: Team admins can now impersonate users for debugging.

Bug Fixes & Improvements

  • FIXEDRLS security: Fixed RLS bypass with proper transaction handling.
  • FIXEDSoft-delete policies: Resolved soft-delete RLS policies not working correctly.
  • FIXEDAPI key decryption: Fixed API key decryption issues.

Title

Features

Features Desc

Pricing

Pricing Desc

Best Tms

Best Tms Desc