Agentโ™ฅ๏ธŽAge
Catalog

AgenticRail Gate - sequence enforcement and verifiable audit receipts for AI agent compliance

Official2 tools

by MSMD-RUA ยท JavaScript

Deterministic runtime enforcement of step order for AI agents: ALLOW/DENY before a step runs.

nz.agenticrail/gate MCP Server

This Model Context Protocol (MCP) server exposes the live AgenticRail enforcement gate to MCP clients as two tools. AgenticRail provides deterministic runtime enforcement of step order for AI agents, allowing only declared steps and refusing replays or skipped steps, then sealing completed sequences with a signed receipt.

๐Ÿ› ๏ธ Key Features

  • Deterministic ALLOW/DENY enforcement before a step runs
  • Enforces declared step order (refuses replays and skipped steps)
  • Signed receipt for each completed sequence
  • MCP adapter over AgenticRail
  • 2 tools

๐Ÿš€ Use Cases

  • Runtime AI agent governance and compliance
  • Audit-trail support for step execution sequences
  • Enforcing step-ordered workflows across MCP clients

โšก Developer Benefits

  • Standard MCP access to the enforcement gate
  • Stateless integration via Streamable HTTP endpoint
  • Works with protocol revision 2026-07-28

โš ๏ธ Limitations

  • Endpoint provided is https://mcp.agenticrail.nz/ (Streamable HTTP, stateless)
  • Protocol revision 2026-07-28 retired the initialize exchange and Mcp-Session-Id

Topics

agentic-aiai-agentsai-governanceaudit-trailcloudflare-workerscomplianceenforcementmcpmcp-servermodel-context-protocolagent-securitypolicy-enforcementruntime-enforcementsequence-enforcement

Related servers

More in Security