Tag

#prompt-injection

AgentsAug 20, 20263 min read

CoSnitch Shows Prompt Injection Is Now A Recon Problem

A named attack coaxed Copilot into mapping its own architecture. One commentator argues the real fix is boring: treat the assistant's context like a network segment.

AgentsJul 13, 20264 min read

OpenAI's Safe Url Defense Targets Agent Prompt Injection

OpenAI details how ChatGPT agents resist prompt injection: a Safe Url check blocks data leaks after a compliance-email scam worked half the time in testing.

AgentsJul 13, 20267 min read

How to Make an MCP Server Read-Only by Construction, Not Prompt

A Redis job-queue MCP server proves 'read-only' in a system prompt is just documentation. Real enforcement means registration logic and ACLs, not vibes.