Below, I will share with you 6 open source projects that have been active on GitHub recently, have commercial application potential and are extremely practical.
Without further ado, let's get down to business.
Copilot open source solution
Since Microsoft announced Microsoft 365 Copilot last week, the word Copilot has been mentioned by people from all walks of life.
" Copilot has caught lightning in a bottle (Copilot has caught lightning in a bottle)".
This is a sentence that Sequoia Capital said not long ago when discussing a new round of technological revolution.
It can be seen that Copilot has begun to become a concrete concept and has gradually entered thousands of households.
In order to allow developers to use the capabilities of the AI Large Language Model (LLM), Microsoft has also open sourced a lightweight SDK on GitHub: semantic-kernel.
This project can be said to be part of Copilot's solution.
The SDK supports and encapsulates multiple design patterns from the latest AI research so that developers can inject sophisticated skills into their applications.
GitHub：https://github.com/microsoft/semantic-kernel
Not only that, the SDK also provides functions such as hint chaining, recursive reasoning, summarization, zero/few-shot learning, contextual memory, long-term memory, embedding, semantic indexing, planning and accessing external knowledge stores as well as internal data.
If you want to try to develop a product that accesses Copilot functions in the future, this project can be used for your reference.
Let the command line also use ChatGPT
In my opinion, the command line terminal is a very important entry point for programmers, and it is also a development tool that everyone uses frequently every day.
An excellent place to greatly increase productivity, ChatGPT should of course be arranged.
Therefore, developers on GitHub have open sourced a very useful command-line tool: Shell GPT .
The tool integrates ChatGPT directly in the command line, through which, it can help you quickly write shell commands, code snippets, Git submission information, documents, comments, etc.
GitHub：https://github.com/TheR1D/shell_gpt
There are mainly the following usages:
- Obtain the execution instructions of a task;
- Directly convert time, distance, weight, temperature and other units;
- Search shell usage in the command line terminal;
- Automatically generate code and store it;
- Talk to ChatGPT on the command line and optimize iteratively.
In simple terms, it is a local command line robot. You give instructions, and it is responsible for executing them. Even if you don’t know how to use the command line to implement a certain task, you can ask it directly.
Like other ChatGPT applications, you need to configure the OpenAI API Key before using Shell GPT.
Feishu GPT
Because many people are now using Feishu to handle tasks such as daily work communication, document collaboration, and video conferencing.
So, not long ago, some domestic developers open sourced a productivity tool on GitHub: Feishu GPT .
GitHub: https://github.com/Leizhenpeng/feishu-chatgpt
It integrates APIs such as Feishu + GPT-3.5 + Dall・E + Whisper, and supports functions such as voice dialogue, role-playing, multi-topic discussion, AI drawing, table analysis, and document export.
In addition to being used for office work, this tool is also good for occasionally fishing.
A code search engine based on GPT-4
The following project is a bit like an open source implementation of the GitHub Copilot X code search function.
Bloop , an open source code search engine.
Answer code-related questions with GPT-4, and search local and remote repositories using natural language, regular expressions, and filtered queries.
GitHub：https://github.com/BloopAI/bloop
The main functions of Bloop are as follows:
- GPT-4 based session search;
- Fast regex search;
- Synchronize local and GitHub repositories;
- Filters that support a variety of different query methods;
- Find functions, variables, or traits using symbolic search
- Precise code navigation with more than 10 popular programming languages.
Since the project code is completely open source, you can also integrate its functions into the internal management system of the enterprise to enhance the retrieval efficiency of technical documents, manuals, tutorials, etc.
ChatPDF open source solution
About two weeks ago, I recommended a project called ChatPDF to everyone .
It can extract the content in PDF files and pass the content to ChatGPT for analysis. After the analysis is complete, you can have a one-on-one conversation with it, as if it is a person who fully understands the content.
However, the project code is not open source, here is an open source implementation: GPT-4 & LangChain .
Using its open functions, you can quickly create a ChatGPT chatbot for your PDF documents.
GitHub: https://github.com/mayooear/gpt4-pdf-chatbot-langchain
The technology stack used includes LangChain, Pinecone, Typescript, Openai, and Next.js.
LangChain is a framework that makes it easier to build scalable AI/LLM applications and chatbots; Pinecone is a vector store for storing PDFs embedded and in text for later retrieval of similar documents.
In order to help everyone better learn the source code of the project, the author also specially recorded a technical principle analysis video with a duration of 23 minutes, which can be viewed in the project README.
VSCode Smart Plugin
It will take a long time before GitHub Copilot X is officially integrated into VSCode.
If you don't want to wait too long, try this open-source VSCode plugin on GitHub: VSCode ChatGPT .
GitHub：https://github.com/gencay/vscode-chatgpt
Its main function is to integrate OpenAI ChatGPT into VSCode, providing functions including but not limited to the following:
- Support GPT-4, GPT-3.5, GPT3 or Codex models;
- Q&A in the sidebar conversation window;
- Stop responding and reduce tokens consumption;
- Generate files or fix codes with one click;
- Export all conversation transcripts in Markdown format;
- edit and repost reminders;
- Code generation via ChatGPT.
Everything is really VSCode...
According to Miller's Law, the human brain can optimally process around 7 chunks of information. Therefore, I will try my best to restrain the open source projects recommended every week, no more and no less, just the right amount for everyone to learn and explore.
At the end of the article, there is still a wave of Amway’s public account: GitHubDaily , usually we mainly share more practical and interesting GitHub developer tools and ChatGPT open source projects, and occasionally talk about recent new things in the technology circle. Interested friends can Pay attention~
All open source projects and tools mentioned in this article have been included in GitHubDaily's list of open source projects.
This list contains many high-quality, interesting and practical open source technology tutorials, developer tools, programming websites, etc. on GitHub.
From 2015 to now, 3500+ open source projects and 21000+ GitHub Stars have been shared accumulatively. If you need it, you can visit the GitHub address below to pick it up:
GitHub：https://github.com/GitHubDaily/GitHubDaily
Reference: 6 amazing ChatGPT projects, open source! - Zhihu if there is any infringement, please contact me to delete.