GPTme: Intelligent Programming Assistant Running in a Command Line Terminal, Localized Alternative to ChatGPT Code Interpreter

General Introduction

GPTMe is a revolutionary terminal AI assistant tool designed to enhance developers' work efficiency. It perfectly combines powerful AI capabilities with the terminal environment, supporting diverse functions such as code execution, file editing, web browsing and visual recognition. As a localized alternative to ChatGPT code interpreter, GPTMe breaks through the limitations of traditional online AI tools without worrying about software restrictions, network access limitations or privacy and security issues. It supports multiple LLM providers including OpenAI and Anthropic, and can also be deployed locally via llama.cpp. GPTMe not only provides a complete end-to-end development experience, but also features intelligent code analysis, file manipulation, web crawling and other functions, which can effectively assist developers in handling all kinds of programming tasks, making it a true all-round AI programming assistant.

Function List

• Terminal Code Execution: Supports running code directly in the local environment via shell and python tools.
• Intelligent file management: Provide file read/write and modification functions, support incremental update using patch tool
• Web Interaction Capabilities: Integration of Playwright for web browsing and information access
• Visual recognition capabilities: can process images, recognize desktop screenshots and analyze web content
• Self-learning optimization: with an output feedback mechanism to continuously improve the assistant's ability
• Multi-model support: compatible with OpenAI, Anthropic and other multiple AI service providers
• Web interface support: provides optional Web UI and REST API interface
• Computer control: supports interaction with GUI applications
• Persistent Agents: support for creating AI agents with persistence
• Intelligent Completion: Provides Tab Completion and highlighting of commands and paths.

Using Help

1. Installation guide

The installation process for GPTMe is very simple and only requires pipx (Python version 3.10+ required):

pipx install gptme

2. Basic use

After the installation is complete, start it by typing the following command directly in the terminal:

gptme

Examples of common commands:

• Create HTML effects:gptme 'write an impressive and colorful particle effect using three.js to particles.html'
• Generate an image:gptme 'render mandelbrot set to mandelbrot.png'
• Optimized configuration:gptme 'suggest improvements to my vimrc'
• Video Processing:gptme 'convert to h265 and adjust the volume' video.mp4
• Code Review:git diff | gptme 'complete the TODOs in this diff'
• Test Fix:make test | gptme 'fix the failing tests'

3. Guidelines for the use of advanced features

3.1 System commands

• /undo: Undo the previous operation
• /log: View dialog log
• /tools: Display the list of available tools
• /edit: Using the editor to modify the dialog
• /rename: Rename the dialog
• /fork: Create a copy of the dialog
• /summarize: Summarize the current dialogue
• /replay: Re-execute the code blocks in the dialog
• /impersonate: Simulation assistant behavior
• /tokens: Shows the number of tokens used
• /export: Export dialogs to HTML format
• /help: Display help information
• /exit: Exit procedure

3.2 Tool integration

GPTMe supports multiple tool integrations:

• Code execution environment: supports Python and Shell commands
• File manipulation tools: support for read, save, append and patch functions
• Browser integration: allows for web access and information capture
• Visual processing: support for image recognition and screenshot analysis
• Version control: integration with GitHub related features

4. Performance optimization recommendations

• Use the workspace option to specify the working directory
• Selection of the appropriate AI model by model parameters
• Enable or disable specific tools as needed
• Use of non-interactive modes to improve processing efficiency