Microsoft has open-sourced its GitHub Copilot Chat extension for Visual Studio Code, making the full source code available on GitHub. The July 1, 2025, release is a key step in its plan for a more transparent and secure AI editor, a vision first detailed in May 2025.
Developers can now inspect the code to see how the popular assistant works. This move aims to build community trust and improve security through public input. The extension, with over 35 million installations, is central to Microsoft’s AI strategy and is available under the MIT license.
A New Milestone for Open-Source AI
The decision to open-source Copilot Chat marks the first major milestone in Microsoft’s publicly stated vision. The company’s goal is to build a more open AI editor, acknowledging that the rapid growth of open-source AI has reduced the need for secrecy around its own tools.
By exposing the extension’s inner workings, Microsoft is betting on transparency to bolster security. With development tools increasingly targeted by malicious actors, a public codebase allows for a crowdsourced defense, where the community can help spot and fix vulnerabilities faster than a closed team.
The Road to an AI Agent: MCP and Feature Expansion
This release is the latest step in Copilot’s rapid evolution from a simple code-completion tool into a sophisticated coding assistant. Over the past year, GitHub has aggressively expanded its capabilities, adding AI-powered code reviews and extending its reach to Apple’s Xcode and the Windows Terminal.
A key moment in this transformation was the June 2025 release of VS Code 1.101. That update integrated the Model Context Protocol (MCP), a standard that allows AI agents to securely interact with developer tools and resources. This protocol provides the essential plumbing for Copilot to function as a true “AI agent.”
The protocol, first introduced by Anthropic, is conceptually similar to the Language Server Protocol (LSP). Just as LSP decoupled programming languages from specific editors, MCP aims to create a universal connector for AI agents and data sources, a move likened to a “USB-C port for AI applications.”
This agent-like capability allows Copilot to handle complex, multi-step tasks autonomously. As described by GitHub’s feature page, it can be assigned an issue, clone a repository into a secure environment, and submit its work as a draft pull request. One developer described the experience as transformative: “Watching the agent tick off its plan in real time felt a bit like peeking over a teammate’s shoulder – except this teammate never gets distracted by Slack.”
Under the Hood: Unveiling System Prompts and Building Trust
For many developers, the most significant aspect of this open-sourcing is the newfound ability to see the extension’s system prompts. These are the master instructions that guide the AI’s behavior, personality, and constraints. This transparency is a boon for prompt engineers.
The repository reveals the specific prompts used for different slash commands, such as `/explain`, `/fix`, and `/test`. This insight demystifies the “magic” behind the AI’s responses and provides a practical education in advanced prompt engineering.
This level of openness directly addresses long-standing questions about data privacy and what contextual information is sent to the large language models. Developers can now verify for themselves what data leaves their machine, a critical factor for building trust, especially within enterprise environments.
A Consolidated and Open Future for Copilot
Microsoft has also confirmed its long-term strategy involves consolidating its Copilot extensions. The functionality of the original, closed-source extension, which provides the well-known inline code completions, will eventually be merged into the new open-source Copilot Chat extension.
This will unify all major AI features into a single, open-source module. The move simplifies the architecture and reinforces Microsoft’s commitment to an open ecosystem. GitHub CEO Thomas Dohmke celebrated the recent VS Code update that paved the way for these changes, stating, “Big news for the vibecoders: There’s a new version of GitHub Copilot live in VS Code 1.101 today, with updates to bring you more MCP support, smarter chat tools, and better source control.”
By embracing an open-source model for its flagship AI developer tool, Microsoft is not only aiming to improve its product but also to position VS Code as the central, trusted hub for the future of AI-assisted software development. The success of this strategy will now hinge on the community’s engagement.
