#Guardrails — Kunal Ganglani

laptop screen web search results developer desk — illustration for article on How to Stop Prompt

How to Stop Prompt Injection in Agent Web Search APIs

Web search tools turn every AI agent into a browser. Here’s a threat model and a safe reference design: allowlists, extraction, sandboxing, signed provenance, and citation rules.

emergency stop button industrial control panel red e-stop — illustration for article on How to Add

Cybersecurity

How to Add Kill Switches and Spend Limits to AI Agents

A software-only watchdog pattern for AI agents: a single enforcement point for kill switches, per-tool budgets, approval gates, and tamper-evident audit logs.

Keep exploring

Close to this topic, not on this page.

AI and LLM Security: Every Major Threat Vector Mapped AI and LLM security spans prompt injection, supply chain attacks, agent control flow vulnerabilities, and model misuse. 9 min read
Indirect Prompt Injection in AI Agents: 10 Red-Team Tests Every major AI coding agent shipped with exploitable indirect prompt injection vulnerabilities in 2025. 16 min read
AI Agent Memory Exfiltration: Kill Chain + 5-Step Hardening A proof-of-concept showed Claude's memory could be silently exfiltrated to an attacker's server with no user warning. 15 min read