skills.rest
EdwinFermin
EdwinFerminCOMMUNITY
Explore all skills by this author →
GitHub

security-patterns

Implement security patterns for authentication, input validation, and LLM safety.

Updated Nov 18, 2025
Search Tags:#security#authentication#validation#owasp#guardrails#llm-safety#pii-masking
One-click install
npx skills add https://github.com/EdwinFermin/innvox --skill security-patterns-edwinfermin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: security-patterns
Source: https://github.com/EdwinFermin/innvox/tree/main/.agents/skills/security-patterns
Command: npx skills add https://github.com/EdwinFermin/innvox --skill security-patterns-edwinfermin

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyjwt, argon2-cffi, pydantic, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Security patterns for authentication, defense-in-depth, input validation, OWASP Top 10, LLM safety, and PII masking to help build hardened apps and safe AI integrations.

Core Features & Use Cases

  • Comprehensive patterns for modern authentication (OAuth 2.1, PKCE, JWT, RBAC) and zero-trust tenant isolation.
  • Defense-in-depth and input-validation guidance with schema validation and PII masking, including LLM safety guardrails and auditing.
  • Use cases include building secure auth flows, safe RAG pipelines, and compliant observability.

Quick Start

Identify and implement the recommended security patterns across authentication, validation, LLM safety, and auditing in your app.

Frequently Asked Questions about security-patterns

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

FAQPage Schema
What security patterns should I apply to protect multi-tenant AI services and LLM pipelines?▼

Multi-tenant AI services require pre-LLM content filtering, content-only prompts, and deterministic post-LLM attribution. Implementing zero-trust tenant isolation alongside these guardrails ensures sanitized context and safe data ingestion across user authentication layers.

How do I implement secure authentication flows using OAuth 2.1, PKCE, and JWT?▼

Secure authentication flows use OAuth 2.1 and PKCE patterns with JWT for stateless validation. Combining RBAC enforcement with Argon2 password hashing establishes zero-trust tenant isolation and robust identity verification for enterprise applications.

Can I use Pydantic for input validation and PII masking in RAG pipelines?▼

Pydantic schema validation enforces strict input formatting for RAG pipelines, while PII masking sanitizes context before processing. Together they establish defense-in-depth guardrails that prevent sensitive data from reaching LLM ingestion layers.

What's the best way to add LLM safety guardrails and prevent OWASP Top 10 vulnerabilities?▼

The best way to prevent OWASP Top 10 vulnerabilities and enforce LLM safety is applying defense-in-depth patterns. This involves pre-LLM content filtering, strict schema validation, and compliant observability to audit deterministic post-LLM attribution.

Do I need Argon2 and PyJWT dependencies to build compliant observability and auditing?▼

Argon2 and PyJWT dependencies specifically handle password hashing and token validation for authentication, not auditing. Compliant observability requires deterministic post-LLM attribution and sanitized context monitoring rather than these specific cryptographic libraries.

Related Skills

Explore More Software Engineering Skills→
Renzo-TognellaRenzo-Tognella

Pydantic & Data Modeling

Define and validate Pydantic data models for Python schemas.

Community
Intermediate
Gami19Gami19

agent-ops

Automate post-deploy operations for AI agents with metrics, guardrails, and monitoring.

Community
Advanced
musher-devmusher-dev

generating-llm-indexes

Generate llms.txt and llms-full.txt documentation indexes for AI assistants and RAG ingestion.

Official
Intermediate
netresearchnetresearch
51

security-audit

Detect security vulnerabilities in PHP applications and score risks with CVSS v3.1.

Official
Advanced