Agent Lsp from Blackwell Systems is a free MCP server designed to connect AI agents with existing Language Server Protocol (LSP) intelligence. Built with Go and distributed as a single Go binary, it orchestrates tools like gopls and rust-analyzer, making their capabilities accessible to AI for coding tasks. This platform enables AI agents to perform complex code analysis, refactoring, and verification by providing structured, type-aware access to development stack insights. It supports various AI clients and offers a foundational layer for advanced automated code manipulation.

What does Agent Lsp connect to?

Agent Lsp functions as a Model Context Protocol (MCP) server, specifically designed to connect with and orchestrate existing Language Server Protocol (LSP) servers. It integrates with popular language servers such as gopls, rust-analyzer, jdtls, pyright, and typescript-language-server. This connection allows Agent Lsp to act as a crucial bridge, translating the rich code intelligence from these LSP servers into a format accessible to AI agents. It effectively makes language-specific insights available for automated coding tasks, enhancing AI agent capabilities across various programming languages.

Core features and tools

Agent Lsp exposes a collection of callable tools, reported to be between 50 and 66 in total, covering diverse programming tasks. These tools are organized into categories including navigation, analysis, refactoring, diagnostics, formatting, speculative execution, composite exploration, safe editing, build, and test. Specific named tools include blast_radius, apply_edit, simulate_edit, get_definition, find_references, and get_tests_for_file. These capabilities provide AI agents with structured access to codebases, enabling precise and informed actions within the development environment.

Who is it for?

Agent Lsp targets AI agents and developers who are building or utilizing them for coding tasks. Its purpose is to equip AI agents with structured, type-aware access to code intelligence, enabling them to perform complex operations like refactoring, safe editing, and verification. This platform is ideal for automating development workflows and enhancing AI-driven code manipulation. Skip this if your primary goal is manual code editing without AI agent integration for advanced development tasks.

How to get started

For the HTTP+SSE transport, access is granted through Bearer token authentication. When using stdio, Agent Lsp operates as a subprocess.

Use cases

• /lsp-refactor: A multi-step process involving impact analysis, speculative preview, applying changes to disk, verifying the build, and running affected tests.
• /lsp-safe-edit: Previews edits in memory, checks for diagnostic differences, and applies the edit only if it is deemed safe.
• /lsp-verify: A three-layer verification process that performs LSP diagnostics, a compiler/build step, and test execution after code edits.

Agent Lsp provides AI agents with a powerful interface to code intelligence, orchestrating LSP servers to enable advanced automation. It offers a rich set of tools and predefined workflows, supporting critical development operations like refactoring and safe code modifications. The platform supports common AI clients and transport protocols, positioning it as a key component for AI-driven development environments.