MCPort
Supports exposing a Workspace to the public internet via Cloudflare Tunnel, with authentication options, for remote MCP access.
Provides read-only Git operations within a selected Workspace, including status, diffs, history, and blame.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@MCPortWhat files did I change in the last commit?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCPort
MCPort is a local-first MCP tool for working with projects through AI. It lets an AI client read code, search files, understand symbols, inspect Git, edit files, and—when explicitly authorized—run controlled development commands inside selected Workspaces.
Project files, runtime state, and credentials stay on your machine by default. You can use MCP locally or enable OAuth/Bearer Token access for selected Workspaces over the public internet.
中文说明:README.zh-CN.md
Community: Linux.do
Screenshots
Workspace status and AI capabilities

Public MCP and OAuth settings

Language server management

Related MCP server: Softority MCP Server
Features
Read and search project files, source code, images, and directories
Code indexing plus definition, reference, Hover, and document-symbol support
Git status, diffs, history, and blame
Contained file creation, editing, import, copy, move, and deletion
Checkpoints, tasks, validation, and recovery for file changes
Controlled command execution with allowlists, timeouts, output limits, and local confirmation
Local or public MCP access for one or more Workspaces
Per-language LSP installation and updates, with custom server support
Optional Computer Use with screen, mouse, and keyboard control gated by per-action approval
Connect an AI client
In MCPort, add a project under Projects, enable MCP for the Workspace, then copy its connection information. Add the copied MCP address to your AI client’s MCP settings. For public access, use the Workspace public address and its configured OAuth or Bearer Token authentication.
Common clients that support MCP include Claude Desktop, Cursor, Windsurf, Cline, Continue, and other MCP-compatible clients. Their settings locations and configuration formats vary; follow the client’s current MCP documentation and paste the address or configuration supplied by MCPort.
After installing or updating an LSP, ask the AI client to call server_info again so it can refresh the available language capabilities.
macOS installation
The current macOS release is unsigned. After dragging MCPort.app to /Applications, run:
xattr -dr com.apple.quarantine "/Applications/MCPort.app"Only do this for the official download after verifying SHA256SUMS.txt.
Connections
Local access accepts loopback requests by default. Public access can use Cloudflare Tunnel, TryCloudflare, FRP Client, or an external tunnel, with OAuth (recommended) or Bearer Token authentication.
Public Workspaces use routes like:
https://mcp.example.com/w/<workspace>/mcpFiles, code, and LSP
File operations remain inside the selected Workspace after realpath and symlink checks. The built-in code index supports TypeScript/JavaScript, Python, Go, Rust, Java, C/C++, and PHP.
Language servers are managed independently and can be installed or updated on demand. Supported languages include TypeScript/JavaScript, HTML, CSS/SCSS/LESS, Python, JSON, YAML, Markdown, Go, Rust, Java, C, C++, and PHP. Custom LSP definitions can be added for other languages.
Permission tiers
readonly: project browsing, search, code understanding, read-only Git, and historystandard: file changes, imports, checkpoints, tasks, and quick validationfull: controlled commands, command sessions, recovery, and complete validation
Commands use an executable name plus an argument array, never an arbitrary shell string. They remain subject to the global command switch, exact allowlist, timeouts, output limits, and high-risk confirmation.
Computer Use is disabled by default and initially available only on local MCP connections in the full tier. A separate switch can expose it to authenticated public Workspaces using the full tier. Screenshots and desktop input still require explicit MCPort Desktop confirmation for every action.
Security
Workspace boundaries are checked after resolving real paths and symlinks. File changes support SHA256 baselines, transactional rollback, and checkpoints. Public Workspaces always require authentication. Tokens, tunnel credentials, and other secrets use system secure storage and are redacted from logs and traces.
The Runtime runs as the logged-in OS user. The command allowlist is not an operating-system sandbox.
Documentation
Contributing
MCPort is maintained by Sky and released under the MIT License. Issues and pull requests are welcome. See CONTRIBUTING.md, SECURITY.md, CODE_OF_CONDUCT.md, privacy notes, and third-party notices.
Third-party LSPs are installed from their own official package managers or release channels and are not bundled with MCPort. Follow each project’s license and terms.
This server cannot be deployed
Maintenance
Related MCP Connectors
Securely search and manage workspace context files for AI agents and teams.
Git-backed platform for skills, tools, and context for AI agents
Cross-agent artifact workspace with provenance across Claude Code, Codex, Cursor, LangGraph.
Give your AI agent a persistent map of your project's structure, dependencies, and bugs.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceProvides secure, sandboxed file system access for AI assistants to read, write, and manage project files with controlled command execution capabilities, all confined to a designated workspace directory.MIT
- AlicenseNot gradedqualityAmaintenanceProvides AI clients with safe, structured access to local filesystem, Git repositories, and project contexts, featuring file operations, git status/diff, project management, and built-in code review/bug analysis prompts.1,140 npmMIT
- FlicenseNot gradedqualityAmaintenanceEnables AI agents to develop within a local project workspace by reading and modifying files, running commands and tests, checking Git state, and persisting progress as history sessions that can be restored in later conversations.-
- FlicenseNot gradedqualityBmaintenanceEnables AI agents to turn local projects into persistent workspaces, providing file editing, command execution, Git integration, and cross-session history management.-