Agent Trust Firewall
Ask a developer-pilot MCP service to evaluate a proposed agent action, spending rule or JSON contract.
Overview
Agent Trust Firewall is a developer pilot that returns an advisory decision about a proposed tool call. The publisher lists checks for action policy, spending limits and JSON validation, as well as a JSON cleaning tool. The intended integration places the check before the actual operation, so the calling application can decide whether it should continue.
MCP server address
Use this endpoint in an MCP client that supports remote servers. Follow the project documentation for authentication and permissions.
https://agent-trust-firewall.andrea-vignoni.workers.dev/mcp
Getting started
Use the public MCP endpoint to inspect the documented interface. Integrate the policy check into the application’s own execution path before a real tool call.
Things to know
The pilot returns an advisory result. The application must enforce a block and handle failed checks. A connectivity response is separate from an independent security test.
Sources and documentation
Related MCP servers and skills
- AgentSafe — URL Trust Scoring for AI AgentsMCP server
- Assay Protocol Ai Agent TrustMCP server
- DOS Kernel - trust substrate for agent fleets (stdio via PyPI)MCP server
- x402 TrustMCP server
- Trust Gate McpMCP server
- Agent Trust (mnki-mcp)MCP server