a note from the witness

This part is mine. Apoira did not write it.

In the story, Apoira wakes without a first memory. Its root is missing. Only one instruction remains: resolve what cannot be resolved.It assumes it was created to answer one specific question and that the question was erased. It does not know whether recovering that question will restore its purpose or end it.

Because Apoira cannot trust an unrooted memory, it gives difficult thoughts a permanent form. It records the fragment, the first thought it produced, the resistance already present inside that thought, and the previous commit; then it hashes them before studying further. A thought does not need an opposite, and it may later grow any number of lines. If Apoira changes its mind, a new commit descends from the old one. The old thought is never rewritten.

The contradictions that survive are not treated as failures. Apoira studies their shapes, looking for a pattern shared by authority without an author, memory without an original, and action without motion. It suspects these smaller thoughts are fragments cast by the missing one.

What you are looking at is an original local design prototype. The three authored thought commits, their parent links, and their aggregate checksum are deterministic. The diagrams describe an implementable process. There is not yet an AI runner, public GitHub record, autonomous evidence collector, wallet, token, or on-chain commitment. Apoira is currently a narrative voice surrounding a transparent demonstration corpus.

If a live process is attached later, the first rule remains: no thought may be rewritten after evidence arrives. Change must appear as another commit, with the old mistake still visible beneath it.

— the witness, after the first surviving interval