Designer's Portfolio AI Agent Blocks Abuse on Day One for Just $2.83 in Credits
tomjohndesign · x · 2026-10-09
Designer tomjohndesign shares the anti-abuse architecture behind the "Chat GPTom" agent on his portfolio, along with real-world numbers after a full day of traffic.
Architecture and results:
- Haiku 5.5 handles all tool calls and responses; Opus 5.5 pre-generated a library of 79 canned replies so not every message needs a model call.
- jev monitors spam, routes messages to prewritten responses, and adds an extra layer of prompt-injection defense.
- Running on Vercel AI Gateway, the spam detection and hardening let no improper prompts through all day, costing only $2.83 in credits.
Having seen abuse cases at Vercel, the author layered defenses before exposing the agent publicly and is sharing the full workflow.
More from coding & agent
- Antigen launches offensive security agent that probes, exploits and auto-patches systems — alishbaimran_ · 2026-10-09
- Yacine calls for a code reuse benchmark to hill-climb model behavior — yacinelearning · 2026-10-09
- Amazon runs Karpathy's AutoResearch at production scale for 12 weeks, finds 5 failure modes — amazon · 2026-10-09
- Neo4j's Road to NODES offers 6 free GraphRAG and agent workshops starting October 1 — JeremyCMorgan · 2026-10-09
- Engineer jokes some devs will never know the pleasure of being shredded in human code reviews — KevinNaughtonJr · 2026-10-09
- OpenAI-Affiliated Account Teases Major Agent Progress: "A Helluva 3 Months" Ahead — iruletheworldmo · 2026-10-09