Fob: Local AI continuity layer: Stop paying the AI context tax. Stay local.

If you are a developer building serious projects with AI right now, you already know the pain of the "AI Context Tax." You spend $2 worth of tokens explaining a bug to Claude Code, get a great answer, and then an hour later you need to jump into ChatGPT or a Cursor session to look at a related issue. Suddenly, you're copy-pasting the same 4 files, re-running the same prompts, and paying for the exact same tokens all over again just to get the next tool caught up. Worse, your actual project files don't remember why a technical decision was made. I built Fob to be the local continuity layer that sits between your project, your AI tools, and your terminal. 💡 What Fob actually does: Durable Project Memory: It stores facts, pins, handoffs, and saved conflict resolutions right in a local folder in your project workspace. Multi-Model Routing: Ask Claude, Codex, both, or run a structured debate/synthesis from one project-aware local dashboard. Local Control Surface: It exposes a headless local API and an MCP server, meaning you can pull your saved project context straight into any compatible AI workflow. Safety First: It provides an explicit approval path. Review diffs and file-edit approvals locally—no agent runs a git command without a typed confirmation. 🛡️ Plus, it includes CLOAK (Free & Open Source) To tackle the other half of the problem—developers accidentally leaking proprietary code to public models just to hit a deadline—I built CLOAK. It’s a free, independent, open-source CLI companion tool that handles AST-based obfuscation and redaction before code ever leaves your laptop. 🚀 The Founder Release Fob is entirely local-first. No database ceremony, no syncing your private repo to a third-party cloud. Your files and your memory stay on your machine. To celebrate the Product Hunt launch, the Founder License is available for a one-time payment of $19, which gets you the Mac build, your license key via Polar, and lifetime access to the local core engine. I’ll be here all day answering questions. Let me know what you think, how your current AI agent setup is fragmenting, and what integrations you want to see next!