11 KiB
11 KiB
# 1.7.2
* Added terminal display, manual command input + AI model via `hyper_terminal` MCP tool
* Fixed bugs
# 1.7.0
* Improved tool compatibility for Gemini
* Fixed the tool selection count error for some MCP tools
* HyperTool, fetch tool will automatically scroll (some data is lazy-loaded and requires scrolling to trigger it), search tool speed enhancement
* Modified variable prefix naming, changing scope to namespace
* Electron version, added settings for startup window size
* Node.js version, added network settings to allow direct connection via local browser
* Fixed bugs
# 1.6.5
* Fixed tool invocation issues for models like Qwen3, added compatibility mode for tool invocation similar to cline, using regex to match <tool_use>
* View repair task results interface, display abnormal issues
* Fixed bugs
# 1.6.3
* Brand new input editor, optimized display, added variable system, supports built-in MCP setting variables, supports JS code variables
* Large model list, supports filtering, optimized display
* Fixed issues with inputting prompts compatible with MCP
* Agent list supports hover descriptions
* WebDAV synchronization optimization
* See MCP supports adding headers
* Chat history performance optimization, using virtual list + display optimization
# 1.5.4
* Fixed issues with web access, added large model test failure problems.
# 1.5.3
* Supports Claude official API
* Optimized built-in terminal, allowing user input
* Supports Agent task invocation failure, using alternative large model
* Added built-in Agent, `MCP helper`, to assist in MCP installation
* Fixed Markdown rendering bug
* Fixed bug for multi-model comparison in chat
# 1.5.0
* Supports chat renaming
* Supports MCP configuration synchronization
* Supports grok3 thinking chain
* Supports AI-generated cron expressions
* Supports message fork cloning
* Supports development mode, quick export of conversation configurations for debugging
* Added enable/disable switch for Claude's MCP
* Added global control enable/disable switch for Task execution
* Optimized tool invocation cancellation, notifying large model users of operation cancellation
# 1.4.17
* Fixed issue of unable to copy in MacOS, and other shortcut keys not working
# 1.4.16
* Supports displaying MCP Server version number and name
* Added detailed error display when model errors occur
* Added detailed error display when MCP errors occur
* Optimized chat history storage, separating conversation messages to reduce loading and synchronization time
* Modified WebDAV synchronization
* Supports MCP in Claude Desktop configuration
* Supports direct viewing of configuration files on the web
* Supports shortcut keys: MACOS `Alt+Cmd+I`, Windows `Ctrl+Shift+I` to open developer tools
* Chat history list supports displaying Agent icons
* Fixed numerous bugs
# 1.4.3
* Rewrote markdown rendering, optimized `Artifacts`, supports opening from the browser, downloading, added support for `Mermaid` pre-release
* Modified tool invocation display
* Supports chat selection of multiple models for comparison
# 1.4.4
* Fixed the default conversation setting MCP invalid issue from the last version
* Fixed that the Agent cannot modify system prompts if there has been no chat
* Fixed the error prompt of no LLM when opening for the first time
# 1.4.1
* Fixed the bug of Gemini tool invocation not supporting multiple tools
* Supports selecting some tools of MCP to save tokens
* Supports quick input with @ + calling Agent
* Fixed bugs
# 1.4.0
* Added built-in MCP tool command line
* Supports changing network access password
* Tool invocation collapsible display
# 1.3.3
* Can confirm and modify large model invocation parameters when invoking tools
* Can click on tools to invoke them for testing
* The knowledge base is planned for redevelopment; currently, it is not recommended for use. You can use OpenAI's embedding model instead, without doing local embedding models
# 1.2.15
* MCP loading progress display, added shortcut button
* Optimized mobile H5 display
* Fixed bugs
# 1.2.12
* Supports asking when invoking tools
* Chat supports displaying time
* Hypetool optimization, settings
* Add historyFilterType hover tip by @xtyuns in https://github.com/BigSweetPotatoStudio/HyperChat/pull/52
# 1.2.2
* Mac, Linux automatically gets PATH, no need to enter PATH
* Added Linux deb package
* Fixed WebDAV first synchronization error
* Fixed bugs
# 1.0.2
* Added temperature setting
* Fixed bugs
* Supports pasting in input box, uploading images
# 0.2.0
* Added knowledge base
# 0.1.1
* Fixed MacOS image upload
* Added quick copy button, supports buttons for copying messages and code
* WebDAV sync function optimization, only saving 10 versions locally
* Markdown code rendering added highlighting
* HTML Artifacts added error capture, supports opening Chrome's console
# 0.1.0
## HyperChat
* Supports two installation methods: plugin market + MCP native installation, quick installation and configuration from the plugin market, welcome to submit plugins, [Github](https://github.com/BigSweetPotatoStudio/HyperChatMCP)
* WebDAV synchronization logic optimized, local backup before synchronization, synchronized every 5 minutes
* LLM added test function, testing whether it supports image input + tool invocation (using a bit of token)
* Chat supports inputting images
* Chat supports displaying images returned by Tool Call from MCP resources
* Fixed bugs
# 0.0.13
## HyperChat
* Optimized the startup speed of MCP Server
* WebDAV synchronization, temporarily does not support MCP configuration synchronization, inconsistent operating systems may cause issues
* Supports setting request method, Stream | Complete
* Supports KaTeX for displaying mathematical formulas
## HyperTools
* Optimized the Tool for opening webpages
# 0.0.11
## HyperChat
* Bot display optimization, supports search and drag-and-drop sorting
* Conversation history supports filtering + searching
* For APIs that do not support counting token consumption for large models, such as Qianwen, support fuzzy statistics by word count
* Input box supports drag-and-drop file quick input
* My LLM Models list supports drag-and-drop sorting
* Supports quickly configuring Ollama and Qwen API models
## HyperTools
* Optimized the Tool for opening webpages, better information extraction. For example, answer this question `https://store.epicgames.com/zh-CN/ 哪个游戏是限时免费的？`