Skip to content
| Marketplace
Sign in
Visual Studio Code>Extension Packs>Copilot RAG for Business DocsNew to Visual Studio Code? Get it now.
Copilot RAG for Business Docs

Copilot RAG for Business Docs

Tadeu Luis Pires Gaudio

|
30 installs
| (0) | Free
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Copilot RAG for Business Docs

Load external content to feed GitHub Copilot with relevant context from external agents or flows.

This extension expands GitHub Copilot's capabilities to work with private business documents in the RAG (Retrieval-Augmented Generation) model. Simply type @findDocs to search for specific documentation for your projects.

Commands

  • @findDocs: Provide your business logic or application rule.

Extension Settings

  • RAG Agent URL: Enter the RAG agent URL.
  • RAG Agent KEY: Enter the RAG agent KEY.

About the Project

Implementing a RAG architecture in VS Code, language models (LLMs), while impressively advanced, are language processing models and therefore need to be fed with up-to-date textual contexts to provide relevant suggestions.

For example, in early 2024, GPT-4 might not correctly write a route handler for Next.js's App Router and might use outdated syntax, as shown below:

Incorrect LLM Response

The solution: provide the model with the appropriate context. This idea is often incorporated into the "RAG" architecture.

RAG Definition

According to IBM's definition:

"Retrieval-augmented generation (RAG) is an AI framework to improve the quality of responses generated by LLMs by grounding the model in external knowledge sources to complement the LLM's internal representation of information."

The Copilot RAG for Business Docs extension aims to implement a basic RAG architecture.

Roadmap

Loading a web page is often not enough to create a suitable prompt. This initial version does not analyze text, so do not expect significant improvements in GitHub Copilot suggestions after loading a web page.

Potential improvements on our roadmap:

  • Allow the use of your own vector repository.
  • Split content into chunks with Langchain utilities and add a query layer with SQLite to build relevant context.
  • Accept an OpenAI API key to use embeddings for proper semantic search.
  • Load GitHub documentation repositories containing Markdown files.
  • Integrate more directly with GitHub Copilot (this part does not depend on us).

Alternatives

The Copilot RAG aims to mimic Cursor's context-loading features, such as @Docs.

Cline is another powerful alternative with contextualization features.

Despite the existing competition, this extension is targeted at GitHub Copilot as it is provided free of charge to open-source developers.

To keep the experience as simple as possible, the Copilot RAG for Business Docs does not require any API key and uses basic retrieval patterns that do not rely on a model. Note: we do not use embeddings and vector search.

Contribute

We use a standard VS Code extension development workflow.

The debugger allows you to test the extension live.

Copilot RAG is developed by Tadeu Luis Pires Gaudio, a Brazilian developer. This extension is not affiliated with GitHub Copilot or Microsoft Copilot.

Release Notes

0.0.1

  • Initial release.

0.0.2

  • Fix README infos.
  • Fix update KEY extension settings.
  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2025 Microsoft