Caliber CLI auto-generates AI configs, tops 150 stars

Caliber's best move is turning agent instructions into a managed artifact instead of a static prompt dump. The local-Llama framing may pull attention on Reddit, but the real product is config hygiene for any team shipping with AI coding assistants.

• –Deterministic offline scoring gives teams a cheap trust layer before an LLM rewrites anything.
• –Backup and undo matter because bad AI instructions are subtle and expensive to debug after the fact.
• –Support for Claude Code, Cursor, and Codex makes this useful as a shared standard, not just an editor-specific helper.
• –The category is crowded with tools like Rulesync and ai-rulez, so Caliber's edge has to come from scoring quality and drift detection, not just file generation.