AI agent tools enable GitHub Copilot and any Model Context Protocol (MCP)-compatible AI agent to perform core Business Central development tasks on your behalf—building, publishing, searching symbols, running diagnostics, and debugging AL extensions—without leaving your editor or workflow.

These tools are designed for two usage surfaces that share a consistent set of tool names and behaviors—the well-known Visual Studio Code experience, and a new MCP for AL, called almcp, that can be used outside of Visual Studio Code—for example, for command line.

Surface	Description	Best For
Visual Studio Code Language Model Tools	Tools registered with the Visual Studio Code Language Model Tools API, available to GitHub Copilot in Agent mode and any Copilot-compatible extension	Developers working interactively in Visual Studio Code
AL MCP Server (`almcp`)	A stand-alone Model Context Protocol server (altool launchmcpserver) that exposes AL tools over STDIO or HTTP	Headless environments, CI/CD pipelines, AI agents that run outside Visual Studio Code

The following agentic tools are available across both surfaces. Some tools are exclusive to one surface due to the nature of the operation.

Tool	Description	Visual Studio Code	AL MCP
`al_build`	Build the AL project and generate the `.app` package	✓	✓
`al_compile`	Validate AL code without generating an `.app` (MCP only)	—	✓
`al_publish`	Publish the extension to Business Central (cloud or on-premises)	✓	✓
`al_downloadsymbols`	Download dependent symbol packages required for compilation	✓	✓
`al_symbolsearch`	Search AL symbols across the project and its dependencies	✓	✓
`al_getdiagnostics`	Retrieve filtered compilation diagnostics	✓	✓
`al_getpackagedependencies`	List the project's `app.json` dependencies (MCP only)	—	✓
`al_debug`	Start a debug session without republishing (Visual Studio Code only)	✓	—
`al_setbreakpoint`	Programmatically add, remove, or toggle breakpoints (Visual Studio Code only)	✓	—
`al_snapshotdebugging`	Manage snapshot debugging sessions (Visual Studio Code only)	✓	—
`al_auth_login`	Authenticate to Microsoft Entra ID (MCP only)	—	✓
`al_auth_logout`	Clear cached authentication tokens (MCP only)	—	✓

These tools, combined with enhanced error handling that provides actionable hints and structured diagnostics via a get_errors tool, mean a coding agent can independently detect and act on compilation failures, iterating on code until it compiles cleanly before proceeding to publish.

Tools also now suggest logical next steps after execution, reducing context switching and allowing agents to chain workflows naturally.

Using tools in Visual Studio Code with GitHub Copilot agent mode

When GitHub Copilot is running in agent mode, it can call AL tools automatically in response to natural-language requests. You can also reference a tool explicitly in Copilot Chat by prefixing its name with #.

As an example, asking "Find all codeunits related to posting" will have GitHub Copilot call alsymbolsearch with appropriate filters as input to provide an overview of related codeunits.

Using the AL MCP server for command-line scenarios

The AL MCP server is a stand-alone process that exposes AL agent tools over the Model Context Protocol. It can be launched with the ALTool. You can configure your MCP-compatible agent (for example, GitHub Copilot CLI, Claude Code, a custom agent built on the MCP SDK, or a CI pipeline agent) to launch the command above as an MCP server and have the agent discover available AL tools automatically through the MCP tools/list call.