security-hardening

// AIDefence security layer with prompt injection blocking, input validation, sandboxed execution, output sanitization, and STRIDE threat modeling.

$ git log --oneline --stat

stars:384

forks:73

updated:March 4, 2026

SKILL.mdreadonly

SKILL.md Frontmatter

namesecurity-hardening

descriptionAIDefence security layer with prompt injection blocking, input validation, sandboxed execution, output sanitization, and STRIDE threat modeling.

allowed-toolsRead, Write, Edit, Bash, Grep, Glob, WebFetch, WebSearch, Agent, AskUserQuestion

Security Hardening

Overview

Multi-layered security audit pipeline implementing the AIDefence architecture. Protects against prompt injection, path traversal, and other attack vectors while ensuring compliance with security best practices.

When to Use

Before deploying code to production
When processing untrusted inputs
Security audits of agent-generated code
Compliance verification (OWASP Top 10, CIS)

AIDefence Layers

Prompt Injection Detection - Pattern + heuristic blocking
Input Validation - Path traversal, type coercion, parameter sanitization
Static Analysis (SAST) - Vulnerability scanning, CWE matching
Sandboxed Execution - Network isolation, filesystem restrictions, resource limits
Output Sanitization - Secrets, PII, injection vector redaction

Security Levels

Level	Layers	Use Case
standard	SAST + validation + sanitization	Routine audits
elevated	+ threat modeling + compliance	Pre-release audits
maximum	+ sandbox + full STRIDE + remediation	Critical systems

Agents Used

agents/security-auditor/ - Vulnerability detection
agents/reviewer/ - Code quality verification

Tool Use

Invoke via babysitter process: methodologies/ruflo/ruflo-security-audit