Dev asks if their agentic upgrade safety plan covers the real attack surface
PitifulAppearance509 · reddit · 2026-11-22
A developer on Reddit is weighing whether to upgrade a simple RAG-based help assistant into a true agentic setup where the LLM queries the RAG, reads from the database, and potentially gains write access.
Worried about new attack vectors, they propose:
- Per-tenant token limits on weekly/monthly usage
- Basic agent rules prompting the model to detect and refuse malicious requests
- Carefully securing tools so they enforce permission keys and tenant isolation
They ask the community whether this is on the right track or if key protections are missing.
More from coding & agent
- Google testing Inbox mode for Antigravity to shift users into approving AI outcomes — testingcatalog · 2026-11-22
- Anthropic publishes Zero Trust security framework for enterprise AI agents — matthew_d_green · 2026-11-22
- Hugging Face's Lewis Tunstall Maps the Road from LLMs to Agent Swarms — _lewtun · 2026-11-22
- Grok Build CLI vs AFK Pilot vs T3 Code: feature comparison across three grok runners — PawelHuryn · 2026-11-22
- A local 8B model made a GGUF quant cheat sheet — all 16 numbers check out — Impressive-Kick293 · 2026-11-22
- GPT-6.2 is hard to steer via system prompt — OpenAI says use it directly in Codex — pvncher · 2026-11-22