1.4 About Security and Trust

MicroTx Workflows provides security and governance through authentication, role-based access control, and configurable guardrails for AI-powered execution.

The workflow server is protected by bearer tokens issued by supported identity providers, and you can control access to workflow and Agentic AI metadata by using roles. Guardrails can evaluate prompts and responses outside the LLM call path, helping reduce unsafe input, prompt injection risks, and undesired generated output. You can use RBAC to control who can create, update, view, and use Agentic AI metadata, including prompt templates, agent profiles, tools, LLM definitions, and MCP server definitions.

Guardrails help protect AI-powered workflow steps from unsafe or undesired input and output. You can configure guardrails to evaluate prompts before they are sent to the LLM and responses after they are returned by the LLM. This allows MicroTx Workflows to block or mask configured content outside the LLM itself, helping reduce risks such as prompt injection or unsafe generated responses.