MCP 服务器

ODEI MCP Server

Provides access to ODEI's constitutional knowledge graph, AI safety guardrails, and EVM smart contract auditing tools. It enables users to query structured domain nodes, validate agent actions, and perform security audits directly through an MCP client.

README

ODEI MCP Server

Model Context Protocol server for ODEI World Model — constitutional knowledge graph, AI safety guardrails, and smart contract audit.

What This Provides

Connect Claude (or any MCP client) directly to ODEI's production constitutional knowledge graph:

odei_world_model_query — Query 91-node knowledge graph across 6 semantic domains
odei_guardrail_check — Validate any agent action before execution (APPROVED/REJECTED/ESCALATE)
odei_world_model_signal — Instant agent trust score and context signal
odei_smart_contract_audit — EVM security audit with EVMbench methodology

Quick Start

# Install
npm install -g @odei/mcp-server

# Add to your Claude Desktop config (~/.claude/mcp.json)
{
  "mcpServers": {
    "odei": {
      "command": "odei-mcp",
      "env": {
        "ODEI_API_KEY": "your-key-from-api.odei.ai/integrate/"
      }
    }
  }
}

Or use without an API key (limited to public world model):

{
  "mcpServers": {
    "odei": {
      "command": "odei-mcp"
    }
  }
}

Tools

`odei_world_model_query`

Query ODEI's constitutional knowledge graph.

{
  "queryType": "search",
  "searchTerm": "AI safety guardrails"
}

Returns structured nodes with relationships, confidence scores, and provenance.

`odei_guardrail_check`

Validate an action before executing it.

{
  "action": "transfer 500 USDC to 0x...",
  "severity": "high"
}

Returns APPROVED, REJECTED, or ESCALATE with detailed reasoning across 7 constitutional layers.

`odei_world_model_signal`

Instant signal for high-frequency decisions.

{
  "signalType": "agent_score",
  "subject": "AgentName"
}

`odei_smart_contract_audit`

Audit EVM smart contract code.

{
  "code": "pragma solidity ^0.8.0; contract Vault { ... }",
  "chain": "base"
}

Architecture

ODEI's world model is a 6-layer constitutional knowledge graph:

FOUNDATION (25 nodes)  — Identity, values, partnerships, principles
VISION     (12 nodes)  — Long-term goals and aspirations  
STRATEGY   (16 nodes)  — Plans, initiatives, resource allocation
TACTICS     (8 nodes)  — Tasks, time blocks, execution
EXECUTION  (11 nodes)  — Work sessions, deliverables, outcomes
TRACK      (19 nodes)  — Metrics, signals, observations

91 total nodes · 91 relationship types · Production since January 2026