You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+5-143Lines changed: 5 additions & 143 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -17,155 +17,17 @@ This innovative AI-based plugin is set to supercharge your development process.
17
17
5.**Code Explanation:** Don't just write code, understand it! DevPilot can explain unfamiliar code snippets, helping you grasp what's happening and learn faster.
18
18
6.**Auto-Comments:** Keep your code clear! DevPilot can automatically add comments to your code, ensuring it's easy to understand and maintain.
19
19
20
-
## Prerequisites
21
-
22
-
1. You will need an OpenAI API key / Codellama API endpoint before you use this plugin.
23
-
2. For use of Codellama, see [deploy Codellama into you local environment](https://github.com/openpilot-hub/codellama-deploy).
24
-
25
-
## Installation
26
-
27
-
Currently, the plugin has version requirements for IntelliJ IDEA. The minimum required version is 2021.2, make sure you have the appropriate version installed before attempting to use the plugin.
28
-
29
-
There are three ways for installing plugins:
30
-
31
-
- IDEA official marketplace - You can get the plugin from [here](https://plugins.jetbrains.com/plugin/23322-devpilot).
32
-
33
-
- Installation via compressed package - select "install from disk" in the plugin page settings, and then import the zip package. The download link for the compressed package is available in [Release page](https://github.com/openpilot-hub/devpilot-intellij/releases).
On the configuration page, you can switch the language of the plugin between multiple choices, as well as select different models and their addresses. The username will be displayed on the page during the conversation.
In the dialog box, you can ask any question and send it by pressing Enter or using the send button. DevPilot will then provide an answer and display it on the current page.
61
-
62
-
Select a block of code, then right-click on the "DevPilot" option on the context menu.You can choose from one of:
Then, the window will automatically open and provide relevant suggestions for fixes. You can modify the code based on the suggestions.
77
-
78
-
## Plugin Functionality
79
-
80
-
DevPilot provides developers with rich capabilities to help them eliminate the tedious work of writing test cases and comments, write more efficient code, and troubleshoot code issues, among other things.
81
-
82
-
### Chat with DevPilot
83
-
84
-
The plugin supports chat, the conversation page can be opened through the right-click menu or the entry on the right sidebar.
You can write questions in the dialogue box, send them by pressing enter or the send button, and OpenAI will provide answers and display them on the current page.
89
-
90
-
91
-
92
-
### Generate test cases
93
-
94
-
Users can generate test cases through our plugin. Users can right-click on a selected method and click "Generate Tests".
Afterward, a window will automatically open and the window will return the relevant test cases. Users can directly copy the test or choose to insert or replace the relevant code at the cursor position. They can even create a new test class file.
Plugin supports the function of generating code comments. Users can right-click and select "Generate comments" after selecting a code block.
105
-
106
-
107
-
108
-
Afterward, the window will automatically open and return the relevant annotation results. The edited code will also automatically show the generated annotations and the original code in a diff format, allowing users to compare and accept the corresponding annotations.
109
-
110
-
### Fix bugs
111
-
112
-
The plugin supports the ability to fix code. Users can right-click on "Fix This" after selecting a code block.
Afterward, the window will automatically open and return the relevant performance test results. Additionally, the optimized code and the original code will be automatically compared using diff in the editor, allowing users to optimize their code based on the results.
The overall capabilities of the plugin are context-dependent, but the context may be polluted, and OpenAI or other large models may have token limitations that result in errors when exceeded. Therefore, we provide the ability to clean up by clicking the clear button in the upper right corner of the window, which will clear the entire session.
At DevPilot, we are constantly working on expanding the capabilities of our AI-powered plugin to meet the evolving needs of our users. Our roadmap is a reflection of our commitment to continually improve and innovate.
149
-
150
-
### Near-Term Goals : integration with local LLM
151
-
152
-
We weill soon support Codellama, an opensource LLM developed by Meta, into DevPilot. This will provide you with the ability to leverage the power of LLM directly in your local development environment, without sending any sensitive data to OpenAI.
153
-
154
-
With Codellama integration, users can expect:
155
-
156
-
- More accurate and context-specific code suggestions, Codellama has a 100k context length, so the suggestions it provides will be based on a deeper understanding of the code's context. This will significantly reduce the time developers spend on figuring out the appropriate code to use.
157
-
- With its fine-tuning on a large codebase, Codellama can outperform ChatGPT in certain cases.
22
+
[DevPilot for JetBrains](https://github.com/openpilot-hub/documentation/blob/main/README_JetBrains_EN.md)
158
23
159
-
### Long-Term Goals: Model as a Service (MaaS) Support
24
+
##RAG
160
25
161
-
Looking ahead, we plan to extend our capabilities further by supporting Model as a Service (MaaS). MaaS provides the ability for users to utilize AI models in a service-based manner, without the need for local deployment or maintenance. This feature will enable seamless access to the latest AI models without bearing the computational costs or complexities of managing these models locally.
0 commit comments