About AgentGate Labs

AgentGate Labs builds runtime authorization and security controls for AI agents, MCP servers and autonomous tool calls.

Authorization for software that can act

AgentGate Labs builds an enforceable control point between AI agents and the tools they use. The product evaluates MCP tool calls before execution, then allows, denies or routes the exact request to human approval while preserving audit evidence.

AgentGate is a SOFTIQA product. It is designed for engineering, platform, security, DevOps and GRC teams adopting coding agents and internal agent workflows without giving them unrestricted access to code, data and infrastructure.

What is available

• A managed control plane and HTTP MCP gateway.
• Per-agent identities, tool discovery, deterministic policy and risk scoring.
• Request-bound human approvals with expiry and replay protection.
• Tenant-scoped audit records and CSV / NDJSON export.
• Docker Compose and Helm deployment paths for customer-operated infrastructure.

How we communicate product scope

Security products should be precise about boundaries. The integration matrix distinguishes implemented protocol paths from client compatibility. The Trust Center documents current controls and limitations. Example data is labeled and is never presented as a customer result.

Contact

For product, deployment or security questions, email [email protected]. Responsible disclosure instructions are available in the Trust Center.