v1.3.2

What's Changed

• prevent the sidepanel composer from sending on Enter while an IME composition is still active.
• refine the model vision toggle thumb spacing so it matches the existing settings toggle style more closely.

New Contributors

@Asunfly made their first contribution in #10

v1.3.1

What's Changed

• Improved the AI response experience with Markdown math rendering and enhanced code blocks that now support one-click copy and fullscreen view.
• Refined page sharing UX for new users by defaulting the current tab to shared on first use, plus a clearer in-panel hint when page sharing is turned off. Also fixed the README Chrome Users badge showing not found.

Congratulations on the launch of Tactus in the store!

Use the data export function in the settings page to import data into the extension store version.
Chrome: https://chromewebstore.google.com/detail/tactus/nmhndbemnmmigjooiafblchoegecmkhh
Firefox: https://addons.mozilla.org/firefox/addon/tactus

v1.3.0

What's Changed

New Features

• Supports Firefox; extensions for Chrome and Firefox are in preparation and will be available in their respective stores soon. Stay tuned!
• Export and import data features are provided to facilitate user migration to the official extensions that will be available later.

Documentation

• update tactus icon!

New Contributors

@hmzf-zhaoyun made their first contribution in #6

v1.2.0

What's Changed

🤖 Multi-Provider Native API Support

• Native Anthropic and Gemini API support
• Intelligent provider type system for seamless switching

🔒 Tab Lock

• Locks active tab during AI responses with glow animation
• Content fetching bound to locked tab context

⚡ UX Improvements

• Preset quick actions for common chat operations #4
• Auto-save config with toast notifications
• API endpoint preview below Base URL input
• Links open in new tabs

v1.1.0

What's Changed

🎨 Vision & Image Support

• Added vision capability settings for models to enable image input
• Implemented direct image paste support in the input box (Ctrl/Cmd+V)
• Updated upload hints to include paste functionality

🔧 API & Configuration Improvements

• Smart Base URL handling: auto-appends /v1/chat/completions or /chat/completions based on trailing slash
• Added HTTP MCP support with Bearer Token and OAuth authentication
• New option to bypass Readability algorithm for specific websites

🚀 Functionality Enhancements

• Added stop button to terminate running agents mid-execution
• New setting to limit maximum tool invocations per agent run
• Configurable maximum character limit for web page extraction
• Improved tab display and added text selection quoting in sidebar

🐛 Bug Fixes

• Fixed duplicate model invocation issue with debug logging

📝 Documentation

• Updated README with latest features and usage instructions

v1.0.3

What's Changed

New Features

• Theme Switching - Support for light mode, dark mode, and system preference auto-detection
• Floating Ball & Text Selection Settings - Added toggle switches for floating ball and text selection quote features
• Message Edit & Copy - Added ability to edit and copy messages in conversations

Bug Fixes

• Fixed tool messages being lost when calling models

Documentation

• Added UI design skills for consistent extension styling
• Moved tactus skills to resources folder

New Contributors

@caulif made their first contribution in #1

v1.0.2

What's Changed

fix: Resolve tool call argument parsing failures caused by incomplete JSON from certain models (e.g., 硅基流动 GLM-4.7) that truncate output when generating nested JSON structures. Add tryFixIncompleteJson function to automatically repair missing closing brackets in tool call arguments while preserving normal JSON parsing behavior.

Tactus v1.0.1

What's Changed

feat: add i18n with auto-detection on initial install

• Add internationalization framework
• Implement language settings
• Add first-run auto-detection

Tactus v1.0.0 - Initial Release

🎉 First Release! 🎉

The first browser AI assistant extension supporting Agent Skills is now officially released!

✨ Core Features

• Agent Skills System – Pioneering in-browser Skills support, enabling the import, management, and execution of standardized skill packages
• OpenAI-Compatible API – Supports models from OpenAI-compatible services
• Multi-Model Management – Configure multiple service providers and switch between models with one click
• Intelligent Page Understanding – Automatically extracts page content and supports referencing selected text
• Chain-of-Thought Display – Shows reasoning processes for models like DeepSeek-R1
• Local Storage – Conversation history and Skills data are stored entirely locally, prioritizing user privacy

首个支持 Agent Skills 的浏览器 AI 助手扩展正式发布！

✨ 核心功能

• Agent Skills 系统 - 首创浏览器端 Skills 支持，可导入、管理和执行符合规范的技能包
• OpenAI 兼容 API - 支持 OpenAI兼容服务模型
• 多模型管理 - 配置多个服务商，一键切换模型
• 智能页面理解 - 自动提取页面内容，支持选中文字引用
• 思维链展示 - 支持 DeepSeek-R1 等模型的推理过程显示
• 本地存储 - 对话历史和 Skills 数据完全本地化，隐私优先