# Managing Guardrails

> Enforcement of safety guardrails, axiom verification, secret scanning, and mutability protections.

- Skill: `gitwalter/managing-guardrails` (Agent Skill)
- Install (CLI): `npx skillmds@latest add gitwalter/managing-guardrails`
- Raw SKILL.md: https://api.skillmd.com/api/skills/gitwalter/managing-guardrails/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: gitwalter (https://skillmd.com/u/gitwalter)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/gitwalter/managing-guardrails

---


# Safety Gating & Mutability Protection

This skill manages internal safety enforcements, checking model inputs/outputs for compliance with core axioms, scanning files for exposed secrets, and protecting system files from unauthorized mutation.

## When to Use
Use this skill when auditing workspace secrets, verifying code alignment with ethical rules, or locking down configuration files.

## Prerequisites
- Conda environment initialized.

## Process

Follow these procedures to scan the workspace and enforce safety enforcements.

### Checking Axioms & Safety
Run the checkers to scan the workspace:
```bash
conda run -p D:\Anaconda\envs\cursor-factory python scripts/guardian/axiom_checker.py
conda run -p D:\Anaconda\envs\cursor-factory python scripts/guardian/secret_scanner.py
```

### Auditing Core Value Mappings
Ensure all codebase elements align with value tables:
```bash
conda run -p D:\Anaconda\envs\cursor-factory python scripts/validation/audit_values.py
```

## Best Practices
- **Strict Compliance**: Reject any change that triggers mutability or safety guardrail warnings.
- **Fail Closed**: If a secret scan fails, block the commit or operation immediately.