configuring-guardrails

Name: configuring-guardrails
Availability: InStock
Author: celigo

Configure Celigo guardrails to validate integration data with PII detection, content moderation, or AI agent evaluation.

3|2|Updated May 20, 2026

Search Tags:#data masking #guardrails #content moderation #pii detection #ai agent evaluation #celigo

Ready to run

npx skills add https://github.com/celigo/ai --skill configuring-guardrails

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: configuring-guardrails
Source: https://github.com/celigo/ai/tree/main/skills/configuring-guardrails
Command: npx skills add https://github.com/celigo/ai --skill configuring-guardrails

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Guardrails prevent unsafe, non-compliant, or sensitive data from flowing through Celigo integrations by validating records before they reach downstream systems.

Core Features & Use Cases

PII detection and masking: Detect sensitive entities like email addresses, SSNs, and credit card numbers, with optional masking so data is redacted before it is delivered.
Content moderation: Block or flag harmful content by category (hate, violence, harassment, self-harm, illicit activity).
AI agent evaluation: Apply custom AI-based checks using an OpenAI-backed agent configuration for tailored business compliance rules.
Flow-wide usage: Use guardrails across flows, APIs, and tools to enforce consistent safety and compliance behavior.

Quick Start

Use the configuring-guardrails skill to create a PII guardrail that scans for email_address and enables masking so sensitive values are redacted before downstream steps run.

Frequently Asked Questions about configuring-guardrails

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

FAQPage Schema

How do I detect and mask PII like email addresses and SSNs in integration flows?▼

PII detection and masking in integration flows is configured by creating a guardrail with a specified type and a list of PII entities (like email_address or SSN). Enabling masking redacts sensitive values before data reaches downstream systems.

What is the best way to moderate harmful content before it reaches downstream APIs?▼

Content moderation blocks or flags harmful content by configuring a guardrail with specific moderation categories such as hate, violence, harassment, self-harm, or illicit activity. This validates records before they reach downstream APIs.

Can I use an OpenAI-backed AI agent to evaluate custom business compliance rules?▼

AI agent evaluation applies custom AI-based checks using an OpenAI-backed agent configuration. This allows tailored business compliance rules to validate data flowing through integrations before it proceeds to downstream steps.

How do I configure a confidence threshold for content moderation or PII detection?▼

A confidence threshold is configured as an optional field when setting up guardrails. It works alongside required type-specific fields like PII entities or moderation categories to determine how strictly data is validated.

Does configuring guardrails require specific fields for each validation type?▼

Configuring guardrails requires a guardrail.type field and type-specific fields: pii.entities, moderation.categories, or aiAgent. Confidence threshold and PII masking behavior are optional settings applied during configuration.

Can I apply guardrails across flows, APIs, and tools for consistent compliance?▼

Guardrails apply across flows, APIs, and tools to enforce consistent safety and compliance behavior. They validate data using PII detection, content moderation, or AI agent evaluation before it reaches downstream systems.