Caliber hits 150-star AI setup milestone

This is the kind of boring automation that quietly becomes indispensable. The real win is not generating files once, it's keeping AI context honest as the codebase changes.

• –It emits `CLAUDE.md`, Cursor rules, `AGENTS.md`, skills, and MCP configs from the actual project structure.
• –The audit, diff, review, and undo flow lowers the trust barrier versus blindly overwriting prompt files.
• –Deterministic, offline scoring and private execution are a strong fit for teams that don't want code or context leaving their machine.
• –Supporting Claude Code, Cursor, and Codex makes it easier to standardize one setup across a mixed-tool team.
• –The 90 merged PRs in the post point to real open-source momentum, not just a solo demo.