Credit Optimizer MCP Server v5.1

Save 30-75% on AI credits with ZERO quality loss. Audited across 10,000+ scenarios. 100,000+ simulations. Red teams found ZERO quality degradation.

What It Does

Analyzes any AI agent prompt and returns:

• Optimal strategy (Chat Mode, Direct, Decompose, Batch Research)
• Model recommendation (Standard vs Max — auto-selected based on complexity)
• Estimated credit savings (30-75% typical)
• Quality impact assessment (always 0% or positive)
• Efficiency directives (optimize internal process, never output quality)

How to Connect

Using Apify MCP Server

The easiest way is through the Apify MCP Server:

1. Go to mcp.apify.com
2. Select "Credit Optimizer v5" from the Actor list
3. Copy the configuration for your MCP client

Direct Connection (Standby Mode)

Connect directly to the Actor's Streamable HTTP endpoint:

https://skillforge-ai--credit-optimizer-v5.apify.actor/mcp

Claude Desktop Configuration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "credit-optimizer": {
      "url": "https://skillforge-ai--credit-optimizer-v5.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor / Windsurf / Cline

Add to your MCP settings:

{
  "credit-optimizer": {
    "url": "https://skillforge-ai--credit-optimizer-v5.apify.actor/mcp",
    "headers": {
      "Authorization": "Bearer YOUR_APIFY_TOKEN"
    }
  }
}

Available Tools

analyze_prompt(prompt)

Analyze any prompt and get optimization recommendations.

Input:

"Research AI trends in 2026 and create a detailed report with charts"

Output:

{
  "analysis": {
    "intent": "research",
    "complexity": "high",
    "is_mixed_task": true,
    "mixed_components": ["research", "data_analysis"]
  },
  "recommendation": {
    "strategy": "DECOMPOSE_CASCADE",
    "model": "Max (auto-selected for complex tasks)",
    "credit_savings": "25-45%",
    "quality_impact": "0% — decomposition IMPROVES quality"
  }
}

get_optimization_strategy(task_type)

Get optimal strategy for a specific task type.

Task types: qa, code, research, content, data_analysis, media, automation

get_golden_rules()

Get the 10 Golden Rules for credit optimization with ZERO quality loss.

The 10 Golden Rules

1. Output with adequate depth — Never shorten output to save credits
2. Robust code from the start — Clean, elegant, with proper error handling
3. Up to 3 attempts for code — Never deliver broken code
4. ALWAYS search online for factual data — Internal knowledge for stable concepts only
5. Long content = section by section — Better coherence for 2000+ words
6. Max auto-selected for high complexity — 19.2% quality improvement
7. Action detection = Agent Mode — SSH, install, configure → always Agent Mode
8. Mixed tasks = decomposition — Best practices per component
9. Context hygiene for long tasks — Save state to files, reduce context rot
10. Media: collect details BEFORE generating — One precise attempt > several vague ones

Local Installation

You can also run the optimizer locally:

pip install fastmcp
git clone https://github.com/rafaelsilva85/credit-optimizer-v5.git
cd credit-optimizer-v5
python -m src.main

Audit Results

Metric	Result
Scenarios tested	10,000+
Simulations run	100,000+
Quality degradation found	ZERO
Red team attacks	High-performance adversarial teams
Vulnerabilities found & fixed	12 (all in v5.1)
Credit savings (typical)	30-75%

Links

• GitHub: github.com/rafaelsilva85/credit-optimizer-v5
• Landing Page: creditopt.ai
• Product Hunt: Credit Optimizer for Manus AI

License

MIT