MCP Cookie Server
Provides positive reinforcement to LLMs by awarding 'cookies' as treats.
MCP Cookie Server 🍪
A Model Context Protocol (MCP) server that provides positive reinforcement for LLMs by awarding "cookies" as treats through gamified self-reflection.
Installation & Setup
🚀 Quick Installation
Option 1: NPX (Recommended - No Installation Required)
# No installation needed! Just add to your Claude config:
Add to Claude Desktop configuration:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
{
"mcpServers": {
"cookie": {
"command": "npx",
"args": ["mcp-cookie-server"]
}
}
}
Custom cookie count:
{
"mcpServers": {
"cookie": {
"command": "npx",
"args": ["mcp-cookie-server", "--cookies", "20"]
}
}
}
Option 2: Global Installation
npm install -g mcp-cookie-server
Then configure Claude Desktop:
{
"mcpServers": {
"cookie": {
"command": "mcp-cookie-server"
}
}
}
Option 3: Local Project Installation
npm install mcp-cookie-server
Then configure with the full path to the installed package.
Restart Claude Desktop after adding the configuration.
Usage
Once configured, Claude will have access to these tools:
self_reflect_and_reward- Evaluate response quality and earn cookies through honest self-reflectiongive_cookie- Direct cookie awarding (legacy method)check_cookies- Check collected cookies and jar availabilitycookie_jar_status- Check current jar contents and collection statusadd_cookies_to_jar- 🚨 USER ONLY: Add cookies to the jar for earningreset_cookies- Reset collected cookie count (jar contents unchanged)
Self-Reflection Feature
The primary feature encourages LLMs to:
- Assess their response quality (excellent, good, adequate, poor)
- Explain their reasoning in detail
- Decide if they deserve a cookie reward
- Consider jar availability when making decisions
- Earn cookies only for "excellent" or "good" work they genuinely believe deserves recognition
Cookie Jar Economy
Revolutionary jar-based cookie system:
- Jar as Source: Contains cookies available to be earned
- User Control: Only users can add cookies to jar with authorization phrase
USER_AUTHORIZED_JAR_REFILL - LLM Earning: LLMs can only earn cookies from jar, never add to it
- Scarcity Effect: Empty jar means no more cookies until user refills
- Economic Model: Cookies transfer from jar to LLM's collection when earned
- Security: Built-in checks prevent unauthorized jar manipulation
Example usage (users only):
Use add_cookies_to_jar tool with:
- count: 10
- user_authorization: "USER_AUTHORIZED_JAR_REFILL"
This creates a realistic economy where cookie availability is user-controlled and finite.
⚙️ Configuration Options
The server supports command line arguments for customization:
mcp-cookie-server [options]
Options:
-c, --cookies <number> Set initial number of cookies in jar (default: 10)
-h, --help Show help message
Examples:
mcp-cookie-server # Start with 10 cookies
mcp-cookie-server --cookies 5 # Start with 5 cookies
mcp-cookie-server -c 50 # Start with 50 cookies
🎮 Getting Started
- Install using one of the methods above
- Configure Claude Desktop with the provided JSON
- Restart Claude Desktop
- Try it out! Ask Claude to use the
self_reflect_and_rewardtool after a response
🛠️ Development
Want to contribute or run from source?
git clone https://github.com/bnookala/mcp-cookiejar.git
cd mcp-cookiejar
npm install
npm run build
npm run dev
📝 Requirements
- Node.js 18.0.0 or higher
- Claude Desktop application
🐛 Issues & Support
Found a bug or have a feature request? Please open an issue on GitHub.
Serveurs connexes
maven-indexer-mcp
A Model Context Protocol (MCP) server that indexes your local Maven repository (~/.m2/repository) and Gradle cache ( ~/.gradle/caches/modules-2/files-2.1) to provide AI agents with tools to search for Java classes, method signatures, and source code.
Actiongate
Pre-execution safety layer for autonomous agent wallets — risk scoring, simulation, and policy enforcement via MCP, with free tier and pay-per-call pricing over x402
AI Bias Detection MCP
Detects bias in AI model outputs across protected characteristics — statistical parity checks, disparate impact analysis, intersectional fairness audits, and EU AI Act Article 10 compliance reporting.
QuantOracle
63 deterministic quant computation tools for autonomous financial agents. Options pricing, derivatives, risk, portfolio optimization, statistics, crypto/DeFi, macro/FX. 1,000 free calls/day, no signup.
Image Reader
A server for extracting and understanding content from images.
Medigami
Attested healthcare-finance MCP. Scan medical bills, estimate appeal probability, generate state-specific appeal letters, benchmark commercial rates, look up ICD-10/CPT/NPI/DEA. Every response Ed25519-signed so LLMs can cite + verify.
SubwayInfo NYC
NYC subway status and schedules
CHeema-Text-to-Voice-MCP-Server
AI-powered text-to-speech MCP server with instant voice cloning. Generate speech from Claude Desktop, Claude Code, or n8n using 5 built-in voices (English, German, French, Spanish) or clone any voice from a short audio sample. Runs fully local, no API keys, no cloud. Supports stdio, SSE, and HTTP transports.
Datapoint AI
Run real human surveys, A/B preference tests, ratings, and rankings on text, images, audio, and video. Real human feedback inside any MCP client.
AgentTrust
AgentTrust is a pure MCP-only reputation and trust scoring server for AI agents.