build

Enforce OpenFirma and Cedar policies for AI agent actions with audit logging.

106|5|Updated Mar 22, 2026
One-click install
npx skills add https://github.com/Firma-AI/openfirma --skill build-firma-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: build
Source: https://github.com/Firma-AI/openfirma/tree/main/.skills/build
Command: npx skills add https://github.com/Firma-AI/openfirma --skill build-firma-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openfirma, cedar, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a runtime enforcement boundary for AI agents, ensuring that only authorized actions are executed, thereby mitigating risks associated with unauthorized or erroneous actions by AI agents.

Core Features & Use Cases

  • Policy Enforcement: Defines and enforces policies for AI agent actions, ensuring compliance with predefined rules.
  • Deterministic Execution: Ensures that all actions are evaluated and executed deterministically, without reliance on external models.
  • Audit Logging: Generates signed audit events for denied actions, providing a record of all decisions made by the enforcement boundary.

Quick Start

Set up OpenFirma to enforce policies for your AI agents by defining policies and configuring the Sidecar to intercept and evaluate agent actions.

Frequently Asked Questions about build

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce policies for AI agent actions at runtime?

You can enforce policies for AI agent actions by setting up OpenFirma to define rules and configuring the Sidecar to intercept and evaluate actions. This creates a runtime enforcement boundary that only executes authorized actions.

What is policy-based enforcement for AI agents?

Policy-based enforcement for AI agents is a runtime boundary that ensures only authorized actions are executed, mitigating risks associated with unauthorized or erroneous actions. It provides deterministic execution without relying on external models.

Do I need Cedar policies to use OpenFirma for AI agent management?

Yes, you need Cedar policies and OpenFirma to use this enforcement boundary. These dependencies are required to define policies and evaluate AI agent actions deterministically.

How does deterministic execution work for AI agent security?

Deterministic execution for AI agent security evaluates and executes actions based strictly on predefined policies without relying on external models. This ensures compliance and generates signed audit events for denied actions.

Can I generate audit logs for denied AI agent actions?

Yes, this policy enforcement approach generates signed audit events for denied actions. This provides a reliable record of all decisions made by the enforcement boundary for compliance tracking.

What are the limitations of using policy-based enforcement for AI agents?

Policy-based enforcement for AI agents requires predefined Cedar policies and OpenFirma setup, meaning actions are strictly limited to authorized rules. It does not rely on external models for evaluation, ensuring deterministic but rigid compliance.