ai-agents-threats

star 0

Modelo de amenazas para agentes de IA — SAIF 2.0 y Google Whitepaper May 2025. Rogue actions, sensitive data disclosure, prompt injection directo/indirecto, misalignment, data exfiltration via tool side-effects, data poisoning. Casos reales — Gemini CLI RCE 2025 (poisoned .env, malicious MCP servers, shell filter bypass, toolDiscoveryCommand backdoor, macOS clipboard trap). Triggers — "rogue actions", "acciones no autorizadas", "prompt injection agente", "indirect prompt injection", "sensitive data disclosure agent", "data exfiltration agent", "agent threat model", "amenazas agente ia", "agent attacks", "agent risks", "vulnerabilidades agente", "hijack agent", "agent security risks", "gemini cli rce", "malicious mcp server", "mcp rce", "toolDiscoveryCommand", "clipboard rce agent", "untrusted workspace agent".

By omespino schedule Updated 5/27/2026

play_arrow Run Skill in Manus View GitHub

Skill instructions (SKILL.md) could not be loaded from local cache or raw GitHub repository.

Install via CLI

npx skills add https://github.com/omespino/ai_config --skill ai-agents-threats

Repository Details

star Stars 0

call_split Forks 0

navigation Branch main

article Path SKILL.md