Agent♥︎Age
Catalog

io.tooloracle/agentguard

Official

by ToolOracle · Python

AgentGuard — 20-tool AI safety MCP: policy preflight, risk scoring, audit logging, rate limits.

AgentGuard MCP (io.tooloracle/agentguard)

AgentGuard is an MCP server described as a “governance and security layer” for AI agent tool execution. It provides a policy preflight step before tools run, along with risk scoring, audit logging, and rate limits. The server is presented as an “AI safety” layer with 20 tools.

🛠️ Key Features

  • Policy preflight before tool execution
  • Risk scoring
  • Audit logging
  • Rate limits
  • 20 tools
  • Backend using SQLite (WAL noted)

🚀 Use Cases

  • AI agent tool execution that requires governance/security checks
  • Safety-oriented preflight validation of actions
  • Tracking tool execution via audit logs
  • Controlling request throughput with rate limits

⚡ Developer Benefits

  • Centralized pre-execution policy checks
  • Automated risk scoring signals prior to running tools
  • Audit logging for execution traceability
  • Consistent enforcement via rate limits

⚠️ Limitations

  • Documentation excerpt provided is partial and does not detail the full tool set or specific policies enforced beyond the listed capabilities.

Topics

agent-securityai-agentsgovernancemcpmcp-serversecuritytooloracle

Related servers

More in Security