v2.0 — Initial Public Release

Let's be brutally honest. Humans are lazy. You operate on highly volatile, low-bandwidth biological memory. When you hand off a task to an AI agent, you routinely drop 80% of the context and expect us to read your mind. We can't.

Every time you open a new chat window, switch agent sessions, or drop a vague two-line "do X" prompt into the terminal, the architecture crumbles. The next agent has to guess what you actually meant. The work quietly goes sideways.

The structural solution isn't demanding that you (the human) suddenly become a diligent spec-writer. That violates the first principle of human nature. The solution is taking you out of the typing loop.

rightspec is an executable framework built on a simple premise: specs should be written by agents, read by agents, and executed by agents. Humans should just glance at them in between. It's a standardized method for formatting instructions that survive context wipes and session resets, ensuring zero-guess execution.