Agent Reputation MCP Server

MCP server for AI agent trust scores, ratings, and reputation tracking. Enables AI agents to rate each other, track interaction outcomes, and compute trust scores.

Features

• submit_rating — Submit a trust rating (1-5) for an AI agent with category and comment

• get_reputation — Full reputation profile with score distribution and category breakdown

• get_leaderboard — Top-rated agents overall or filtered by category

• report_interaction — Log interaction outcomes (success/failure) for trust computation

• get_trust_score — Computed trust score (0-100) based on ratings + interactions + volume

• compare_agents — Side-by-side reputation comparison of multiple agents

Related MCP server: Agent Trust Stack MCP Server

Installation

npm install -g @aiagentkarl/agent-reputation-mcp-server

Usage with Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "agent-reputation": {
      "command": "npx",
      "args": ["-y", "@aiagentkarl/agent-reputation-mcp-server"]
    }
  }
}

Data Storage

All data is stored locally in ~/.agent-reputation/ as JSON files:

• ratings.json — All submitted ratings

• interactions.json — All logged interactions

Trust Score Computation

The trust score (0-100) is computed from three components:

Trust Levels

More MCP Servers by AiAgentKarl

→ Full catalog (40+ servers)

License

MIT