⚡ Codeffect

AI-Powered Code Optimization MCP Server

10 AI agents analyze, fix, and verify your code — automatically.

Quick Start • Features • 12 Tools • Pricing • Demo

---

AI writes 41% of all code now. But AI-generated code has 2.74x more security vulnerabilities. Codeffect fixes that — automatically.

What is Codeffect?

Codeffect is an MCP server that optimizes your code through a team of 10 specialized AI agents:

Your Code → [4 Analyzers] → [4 Fixers] → [2 QA Checkers] → Optimized Code

One command. Zero manual review. Works with Claude Code, Cursor, Cline, Windsurf — any MCP client.

Quick Start

# Install
pip install codeffect

# Add to Claude Code
claude mcp add codeffect -- python -m codeffect

# Or add to claude_desktop_config.json
{
  "mcpServers": {
    "codeffect": {
      "command": "python",
      "args": ["-m", "codeffect"]
    }
  }
}

Then just ask Claude:

"Optimize my code" → runs all 10 agents automatically

Features

Detection Team (4 Agents)

Agent	Detects
CodeAnalyzer	Duplicate code, dead code, unused imports, mutable defaults, empty except blocks
PerfAnalyzer	O(n²) loops, string concat in loops, N+1 queries, blocking I/O, memory leaks
CostAnalyzer	API calls in loops, missing timeouts, SELECT *, no caching, token waste
SecurityAnalyzer	SQL injection, XSS, hardcoded secrets, eval(), pickle, weak crypto, path traversal

Fix Team (4 Agents)

Agent	Fixes
DuplicateFixer	Removes unused imports, fixes var→const, cleans commented code
PerfFixer	list→generator, time.sleep→asyncio.sleep, structuredClone()
CostFixer	Adds timeouts, connection pooling, caching, streaming
SecurityFixer	eval→literal_eval, HTTPS, env vars, yaml.safe_load, parameterized queries

QA Team (2 Agents)

Agent	Validates
PreCheckQA	Syntax, function preservation, import safety, no new vulnerabilities
PostCheckQA	Diff generation, metrics comparison, improvement scoring, grade report

12 MCP Tools

Tool	Description
analyze_code	Scan for duplicates, dead code, unused dependencies
analyze_performance	Detect O(n²), memory leaks, bottlenecks
analyze_cost	Find expensive API/DB/token patterns
analyze_security	OWASP Top 10 vulnerability detection
full_analysis	Run all 4 analyzers at once
fix_duplicates	Auto-fix code quality issues
fix_performance	Auto-fix performance issues
fix_cost	Auto-fix cost issues
fix_security	Auto-fix security vulnerabilities
full_fix	Run all 4 fixers in optimal order
pre_check / post_check	QA validation before and after fixes
optimize	One-click full pipeline: Analyze → Fix → QA → Report

Demo

# Before: 18 issues detected
import os
import json
import unused_lib          # ← unused import

password = 'supersecret'    # ← hardcoded secret

def fetch_users():
    for uid in ids:
        requests.get(f'http://api.com/{uid}')  # ← API in loop, no timeout, HTTP
        data = eval(resp.text)                  # ← eval() injection
    cursor.execute(f"SELECT * FROM users WHERE id = {uid}")  # ← SQL injection

# After: Codeffect optimized
import os
import json

password = os.environ.get("PASSWORD")

def fetch_users():
    session = requests.Session()
    # Batch API calls instead of loop
    session.get('https://api.com/batch', timeout=30)
    data = ast.literal_eval(resp.text)
    cursor.execute("SELECT id, name FROM users WHERE id = ?", (uid,))

Result: Score 50 → 85 | 6 security fixes | 4 performance fixes | 3 cost fixes

Supported Languages

Language	Analysis	Auto-Fix
Python	Full (AST + regex)	Full
JavaScript/TypeScript	Full (regex)	Full
Java, Go, Rust, Ruby, PHP	Generic patterns	Annotations

Pricing

Plan	Price	Includes
Free	$0	50 optimizations/month
Pro	$8/mo	Unlimited optimizations, all 15 tools, priority support
Team	$29/mo	Everything in Pro + team dashboard, CI/CD integration

Pay with Crypto (MetaMask)

Wallet: 0x2Fd28D0e56798796175C6AC9cdf1CE5B44670702
Accepted: ETH, USDT, USDC
Networks: Ethereum, Base (low gas), Arbitrum (low gas)

Just ask Claude: "I want to upgrade to Pro" → get payment instructions → send crypto → activate with tx hash.

Why Codeffect?

• For Vibe Coders: You don't need to understand the code to optimize it. Just run optimize.
• For AI Agents: Built as MCP — any AI agent can use it automatically.
• For Teams: Catch issues before they hit production. Save 30-60% on API/DB costs.
• 10 agents > 1 linter: Static analysis + dynamic fixes + QA verification in one pipeline.

Architecture

┌─────────────────────────────────────────────┐
│                 MCP Client                   │
│        (Claude / Cursor / Cline)             │
└──────────────────┬──────────────────────────┘
                   │
┌──────────────────▼──────────────────────────┐
│            Codeffect MCP Server              │
├─────────────┬────────────┬──────────────────┤
│  Detection  │    Fix     │       QA         │
│  ┌────────┐ │ ┌────────┐ │ ┌──────────────┐ │
│  │Code    │ │ │Dup Fix │ │ │PreCheck QA   │ │
│  │Perf    │ │ │Perf Fix│ │ │PostCheck QA  │ │
│  │Cost    │ │ │Cost Fix│ │ └──────────────┘ │
│  │Security│ │ │Sec Fix │ │                  │
│  └────────┘ │ └────────┘ │                  │
└─────────────┴────────────┴──────────────────┘

License

MIT

---

Built for the vibe coding era.
AI writes the code. Codeffect makes it production-ready.