Skip to content

Latest commit

 

History

History
82 lines (54 loc) · 8.12 KB

CONTRIBUTING.md

File metadata and controls

82 lines (54 loc) · 8.12 KB
Raw

Note: This file has been modified from its original contents. pearai-submodule is a fork of Continue (https://github.com/continuedev/continue).

🍐 Contributing to PearAI

🌐 For how to setup, build, and run PearAI, please visit PearAI-App Contributing for the main guide!

📚 PearAI Submodule Specific Guides:

✨ Writing Slash Commands

The slash command interface, defined in core/index.d.ts, requires you to define:

  • name: The text that will be typed to invoke the command.
  • description: The text that will be shown in the slash command menu.
  • run: An async generator function that will be called when the command is invoked. It yields the content to be displayed in the chat and is passed a PearAISDK object for interacting with the IDE, calling the LLM, and accessing chat history.
export interface SlashCommand {
  name: string;
  description: string;
  params?: { [key: string]: any };
  run: (sdk: PearAISDK) => AsyncGenerator<string | undefined>;
}

There are many examples of slash commands in core/commands/slash that we recommend borrowing from. Once you've created your new SlashCommand in this folder, be sure to complete the following 🌟:

  • ✅ Add your command to the array in core/commands/slash/index.ts
  • 🔧 Add your command to the list in config_schema.json. This ensures that Intellisense shows users what commands are available for your provider when they are editing config.json. If there are any parameters that your command accepts, you should also follow existing examples in adding them to the JSON Schema.

📜 Writing Context Providers

A ContextProvider is a PearAI plugin that lets you type '@' to quickly select documents as context for the language model. The IContextProvider interface is defined in core/index.d.ts, but all built-in context providers extend BaseContextProvider.

Before defining your context provider, determine which "type" you want to create. The "query" type will show a small text input when selected, giving the user a chance to enter something like a Google search query for the GoogleContextProvider. The "submenu" type will open up a submenu of items that can be searched through and selected. Examples are the GitHubIssuesContextProvider and the DocsContextProvider. The "normal" type will just immediately add the context item. Examples include the DiffContextProvider and the OpenFilesContextProvider.

After you've written your context provider, make sure to complete the following 🌟:

  • ✅ Add it to the array of context providers in core/context/providers/index.ts
  • 🔧 Add it to the ContextProviderName type in core/index.d.ts
  • 📄 Add it to the list in config_schema.json. If there are any parameters that your context provider accepts, you should also follow existing examples in adding them to the JSON Schema.

🤖 Adding an LLM Provider

PearAI has support for more than a dozen different LLM "providers", making it easy to use models running on OpenAI, Ollama, Together, LM Studio, and more. You can find all of the existing providers here, and if you see one missing, you can add it with the following steps:

  1. 🆕 Create a new file in the core/llm/llms directory. The name of the file should be the name of the provider, and it should export a class that extends BaseLLM. This class should contain the following minimal implementation. We recommend viewing pre-existing providers for more details. The LlamaCpp Provider is a good simple example.
  • 📛 providerName - the identifier for your provider
  • 📤 At least one of _streamComplete or _streamChat - This is the function that makes the request to the API and returns the streamed response. You only need to implement one because PearAI can automatically convert between "chat" and "raw completion".
  1. ✅ Add your provider to the LLMs array in core/llm/llms/index.ts.
  2. 🖼️ If your provider supports images, add it to the PROVIDER_SUPPORTS_IMAGES array in core/llm/index.ts.
  3. 🔧 Add the necessary JSON Schema types to config_schema.json. This ensures that Intellisense shows users what options are available for your provider when they are editing config.json.
  4. 📄 Add a documentation page for your provider in docs/docs/reference/Model Providers. This should show an example of configuring your provider in config.json and explain what options are available.

🧠 Adding Models

While any model that works with a supported provider can be used with PearAI, we keep a list of recommended models that can be automatically configured from the UI or config.json. The following files should be updated when adding a model:

  • 📂 Prompt Templates - In this file you'll find the autodetectTemplateType function. Make sure that for the model name you just added, this function returns the correct template type. This is assuming that the chat template for that model is already built in PearAI. If not, you will have to add the template type and corresponding edit and chat templates.

📖 Adding Pre-indexed Documentation

PearAI's @docs context provider lets you easily reference entire documentation sites and then uses embeddings to add the most relevant pages to context. To make the experience as smooth as possible, we pre-index many of the most popular documentation sites. If you'd like to add new documentation to this list, just add an object to the list in preIndexedDocs.ts. startUrl is where the crawler will start and rootUrl will filter out any pages not on that site and under the path of rootUrl.

⚙️ PearAI Architecture

PearAI consists of 2 parts that are split so that it can be extended to work in other IDEs as easily as possible:

  1. 🖥️ PearAI GUI - The PearAI GUI is a React application that gives the user control over PearAI. It displays the current chat history, allows the user to ask questions, invoke slash commands, and use context providers. The GUI also handles most state and holds as much of the logic as possible so that it can be reused between IDEs.

  2. 🔌 PearAI Extension - The PearAI Extension is a plugin for the IDE which implements the IDE Interface. This allows the GUI to request information from or actions to be taken within the IDE. This same interface is used regardless of IDE. The first PearAI extensions we have built are for VS Code and JetBrains, but we plan to build clients for other IDEs in the future. The IDE Client must 1. implement IDE Interface, as is done here for VS Code and 2. display the PearAI GUI in a sidebar, like here.

🧩 PearAI VS Code Extension

The starting point for the VS Code extension is activate.ts. The activateExtension function here will register all commands and load the PearAI GUI in the sidebar of the IDE as a webview.