Negative Star Innovators

@negativestarinnovators.com

Agent Redactor: Prevent your AI agents from sending sensitive & private data to AI providers. 100% local Opensource MIT License. agentredactor.negativestarinnovators.com Paste Redactor: redactor.negativestarinnovators.com Blog: negativestarinnovators.com

Agent Redactor is an AI Firewall for your sensitive data. It's a proxy that sits between your machine and the AI provider redacting sensitive information using regex, keywords and local AI. Keep your secrets from ever leaving your machine. Free & open source agentredactor.negativestarinnovators.com

Agent Redactor | Local PII Redaction Proxy for AI Coding Agents

Agent Redactor is a free, open-source (MIT) Windows app that sits between your AI coding agents and LLM APIs, redacting PII and secrets locally before they ever leave your machine.

agentredactor.negativestarinnovators.com

We are entering a new witch hunt era but for AI generated text. When your opinions disagree with the majority you more likely to get labeled as AI generated even if its not. Enough people classify you as a witch and you get burnt by receiving nasty comments, downvotes and possible social media bans.

Deciding what to work on each day on the day only works when your mental state is clear. This method fails miserably when your brain is sick (less sleep, illness, etc). Thus plan when your mind is clear for days when your brain is sick. Following premade decisions is easier than making decisions.

AI learns on problems which no longer exists. It would not choose a framework with more boilerplate code even though it met our needs better. It said it would be 'harder' to do. But AI is great at boilerplate code! Question everything. Make your own judgements. AI does not replace thinking.

When data is easily shifted from one platform to another does vendor lock in actually exist any more as a viable strategy for companies? We think AI is eliminating data based vendor lock in. The implication is we should grab the best deal for us without worrying about data based vendor lock in.

Within one week we had been permanently banned from Threads, shadow banned from Reddit and banned and unbanned from X. All accounts 1+ years old. Each company has their own unethical practices. It made us think more carefully about social media. We will discuss each platform in future posts.

The more coding with AI I do, the more I see myself as a facilitator. The questions in my mind these days are: how much information does this coding agent need to complete its task, what access should it have, what irrelevant information could confuse it, and how can I explain myself more clearly?