ThatGuyMarkJames

AI Code Optimizer — AI skill for Claude Code

AI community

It's a plug-in you install into any project once.

How to install AI Code Optimizer

This entry records only its repository, not the path inside it, so there is no exact command to give. Open ThatGuyMarkJames/AI_CODE_OPTIMIZER and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Code Optimizer does

It's a plug-in you install into any project once. After that, every AI agent you use — Cursor, Claude, Antigravity, Windsurf — they automatically clean up the code they write and drop notes right inside your project folder.

Alternatives in AI

Skilldock — SkillDock is an AI skill manager and skill management desktop app for Claude Code, Cursor, Codex, Windsurf, Ge 490 ★
AI Marketing Skills — The marketing fundamentals that make AI actually make you money, written in my own voice, as a ready-to-instal 226 ★
Engrim — The Universal Cross-Model Episodic Memory Standard 219 ★

README

CodeStill — an AI code optimizer, as a web app and an MCP server

**The problem:** AI coding agents ("vibe coding") tend to produce code that *works* but is more complicated than it needs to be — extra wrapper classes, unnecessary abstraction layers, defensive code for cases that can't happen, config systems for values that never change. The result is code the person who "wrote" it (by prompting for it) often can't explain.

**What this project does:** takes that code and produces a simpler version that does the same thing, plus a plain-English explanation of how the code works. It's built two ways at once:

as a normal web app you paste code into, and
as an MCP server, so a coding agent (Claude Code, Claude Desktop, or any MCP client) can call it on its own output during a coding session, before the bloated version ever reaches you.

Tech stack

Layer	Choice	Why
LLM	Groq, Llama 3.3 70B	Free-tier friendly, very fast responses (important for a live demo), good at code
Backend	FastAPI	Minimal boilerplate, async-friendly, automatic `/docs` for free
Frontend	Plain HTML/CSS/JS	No build step — one `uvicorn` command runs the whole thing
Agent integration	MCP (`mcp` Python SDK, `FastMCP`)	The current standard for letting an AI coding agent call an external tool mid-session

Architecture

                    ┌─────────────────────────┐
                    │   core/optimizer.py      │
                    │   (the only place that   │
                    │    talks to the LLM)     │
                    └────────────┬─────────────┘
                                 │
                 ┌───────────────┴───────────────┐
                 │                                │
      ┌──────────▼──────────┐          ┌──────────▼──────────┐
      │ backend/main.py      │          │ mcp_server/server.py │
      │ FastAPI REST API     │          │ MCP tools:            │
      │ POST /api/optimize