CodePortal MCP Server – Overview

CodePortal MCP is a lightweight, local server that bridges the gap between AI assistants and real‑world codebases. By exposing a simple Model Context Protocol (MCP) interface, it lets developers and AI agents like Claude or GPT‑4 quickly organize raw code files into fully structured projects, while also providing secure, on‑premise access to popular generative models such as OpenAI and Google. The server runs entirely on the user’s machine, ensuring that sensitive API keys never leave local storage and that all communication is confined to unless explicitly enabled for external access.

The core problem CodePortal solves is the friction that arises when an AI assistant needs to generate, modify, or refactor code across multiple files. Without a well‑defined project structure, the assistant can’t reliably understand module relationships or maintain consistent naming conventions. CodePortal automatically arranges submitted files into a coherent directory tree, creates necessary files for Python packages, and generates a minimal . This not only speeds up the development cycle but also guarantees that generated code can be run immediately without additional setup. For teams, it provides a single point of entry where AI‑generated code is vetted and organized before being merged into version control.

Key capabilities include:

• Project Organization – Accepts a JSON payload of file names and contents, then builds a fully functional project layout based on the user’s template.
• Secure API Key Management – Stores OpenAI and Google Generative AI keys locally; the server never exposes them to external networks.
• Direct AI Access – Offers a web UI and MCP endpoints that let an assistant call the chosen provider directly, returning text or code snippets in real time.
• Auto‑Shutdown – Prevents lingering processes by terminating the server after 15 minutes of inactivity, keeping the host system clean.
• Optional Public API Mode – Advanced users can expose a secure token‑protected endpoint, enabling remote AI services to submit files or request generation without compromising local credentials.

Typical use cases span from rapid prototyping—where a developer hands a prompt to an assistant and instantly receives a runnable project—to continuous integration pipelines that trigger CodePortal to generate test scaffolding or documentation. In research environments, the server allows AI models to experiment with code transformations in isolation before committing changes. Because all interactions are local by default, latency is minimal and privacy concerns are mitigated.

Integration into AI workflows is straightforward: an assistant can call the endpoint to scaffold a new repository, then use (or the web UI) to ask for function implementations, bug fixes, or documentation. The resulting project can be pulled into a Git workflow, ensuring that AI contributions are versioned and auditable. CodePortal’s design prioritizes developer autonomy—no cloud dependencies, no shared key storage—while still delivering the flexibility of state‑of‑the‑art generative models.