Skip to content
Back to skills

Ground And Guardrail

ASecurity

Add citations, refuse-on-empty-retrieval, and context-constraint to cut hallucination. Reach for this on a faithfulness question.

  • 7 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 23, 2026
ai-agents

Security analysis

A100/100

Scanned September 23, 2026

npx -y skills add mcorbett51090/RavenClaude --skill ground-and-guardrail --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ground And Guardrail?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ground And Guardrail
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/mcorbett51090-ground-and-guardrail/badge)](https://www.skillsdirectory.com/skills/mcorbett51090-ground-and-guardrail)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ground-and-guardrail
description: "Add citations, refuse-on-empty-retrieval, and context-constraint to cut hallucination. Reach for this on a faithfulness question."
---

# Skill: Ground and guardrail

An un-grounded, un-cited generation invites confident fabrication (§3 #7).

## Step 1 — Require citations
Make the model cite the retrieved passage it grounded on (§3 #7).

## Step 2 — Refuse on empty retrieval
If retrieval returns nothing relevant, refuse rather than fabricate (§3 #1 #7).

## Step 3 — Constrain to context
Instruct and check that the answer stays within retrieved context (§3 #7).

## Step 4 — Verify with the eval
Citations are how faithfulness is measured — close the loop (§3 #3 #7).

## Output
A grounding + guardrail design with citations, refusal, and a faithfulness check.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…