I built a 0.07ms Python firewall to stop AI hallucinations
Deterministic policy gates beat LLM guardrails when your agent tries to DROP TABLE.
A collection of small chrome extensions to defend and protest against LinkedIn spying behaviour on chrome based browsers.
Data poisoning script is clever protest, but uBlock Origin handles most blocking already.
Privacy-conscious LinkedIn users, security researchers
uBlock Origin · Privacy Badger · Trace
Hope this helps other worried users around
Deterministic policy gates beat LLM guardrails when your agent tries to DROP TABLE.
Automates LinkedIn spam, violates ToS, ships as obvious proof-of-concept toy.
Ed25519 provenance plus Bayesian trust scoring stops published NeurIPS memory poisoning attacks zero-dep.
Checks if GPTBot and ClaudeBot are blocked before Google even sees your page.
Three-agent AI job matcher when LinkedIn's own AI already fails at this.
OWASP-backed security layer specifically for AI agent memory poisoning risks.