domains / families / security / ai-security-hardening

ai-security-hardening

combined

Harden AI/LLM deployments against prompt injection, data exfiltration, model theft, and supply chain attacks. Covers input validation, output filtering, access control, model API security, and compliance controls for production AI systems. Harden AI/LLM deployments against prompt injection, data exfiltration,

When to use

• Deploying an LLM-powered application handling sensitive user data
• Protecting against prompt injection attacks in AI agents
• Implementing output filtering and content moderation
• Securing model weights and API endpoints from theft
• Achieving SOC2 or ISO 27001 compliance for AI systems

When not to use

• Apply guidance only within authorized scope; test destructive steps in non-production first.
• Docs-only import: upstream scripts and templates not bundled.

Skill page

Conflict

The second copy is a mirror of the first (it cites BagelHole as the source) but adds material constraints the first lacks: it explicitly warns that regex injection detectors are not a security boundary, requires server-side tool/data authorization independent of model text, and notes that the in-process rate limiter is not distributed and must be replaced in multi-worker production. The first copy presents the same code without those caveats. Crowning the first would hide the safer operational guidance, so the copies should be combined.