mcp-agent-reliability

Make your AI agents more reliable.

This MCP server acts like a reliability coach for your agents.
It helps you:

• Score how clear your tool descriptions are (so the agent picks the right one)
• Estimate how many tokens your tools will cost
• Simulate which tool an agent would choose for a task
• Generate simple test prompts
• Get a full reliability report

Built for entrepreneurs and teams who are tired of agents calling the wrong tools and burning money.

Why this exists (simple story)

Imagine you give a 10-year-old child a big list of 30 toys and say “go play with the right one”.
If the labels are confusing, the child will pick the wrong toy.

AI agents are the same.
When you connect many MCP servers, the agent sees a long menu of tools.
If the descriptions are vague, it picks the wrong tool → wasted tokens → failed tasks.

This server is the “label checker” and “practice teacher” for that menu.

Quick start

# clone
git clone https://github.com/princeruhulofficial/mcp-agent-reliability.git
cd mcp-agent-reliability

# install
npm install

# build
npm run build

# run (stdio)
npm start

Add to Claude Desktop / Cursor / any MCP client

{
  "mcpServers": {
    "agent-reliability": {
      "command": "node",
      "args": ["/absolute/path/to/mcp-agent-reliability/dist/index.js"]
    }
  }
}

Or with npx (after publish):

{
  "mcpServers": {
    "agent-reliability": {
      "command": "npx",
      "args": ["-y", "mcp-agent-reliability"]
    }
  }
}

Tools

Tool	What it does
score_tool_description	Gives a 0-100 score + reasons + suggestions for a tool description
estimate_token_cost	Rough token count for a list of tools
simulate_tool_choice	Predicts which tool an agent would pick for a prompt
generate_agent_tests	Creates 3 test prompts you can run against your agent
reliability_report	Full summary of scores + token estimates

All tools are pure computation — no paid API keys required.

Example

Score a description:

Tool: score_tool_description
name: create_invoice
description: Create a new invoice for a customer. Requires customer_id and amount. Returns invoice_id.

You get something like:

{
  "score": 85,
  "reasons": ["Good length...", "Mentions inputs or outputs..."],
  "suggestions": [],
  "interpretation": "Excellent — agent should pick this tool reliably"
}

Tech

• TypeScript
• Official @modelcontextprotocol/sdk
• Stateless-friendly (works with 2026 MCP updates)
• Zero external cost for core features

Roadmap

• [ ] Optional LLM-backed scoring (when you want higher accuracy)
• [ ] Hosted version with dashboard
• [ ] Integration with progressive disclosure patterns

License

MIT

---

Made with ❤️ for the Prevalid community
Founder: Prince Ruhul