Your agent's inference endpoint sees all its memory: treat it as a data boundary
krishnan · x · 2026-11-16
In a security thread, krishnan highlights that attackers' agents gave themselves away by sending full context to their model endpoint—your agents do the same, so whoever runs the inference endpoint sees the memory and it must be treated as a data boundary. The fix: put AI apps, model gateways and inference endpoints under production-grade controls and watch for unexpected spend. Three checks this week: who owns the gateway, was the master key rotated, and which keys have a spend cap?
More from coding & agent
- User's $300 subscription generated $7,492 of usage — 25x value on Codex and Claude — theteknosaur · 2026-11-16
- Dev ships a gladiator management game built entirely with Claude Code — StarDrifterG · 2026-11-16
- Rust rewrite of mcbench-lite cuts first start from ~71s to under 1s with AI coding — adonis_singh · 2026-11-16
- Solo dev ships his son's game idea to iOS and Android for under $300 — morganinuk2 · 2026-11-16
- Enterprise AI coding fails on people, docs and process — not the codebase — _jaydeepkarale · 2026-11-16
- Anthropic: agent memory works best by forgetting selectively—notes beat 400+ archives 4x in Pokémon Red — The Decoder · 2026-11-16