Vision: dna/product/vision-spec-machine.md (v4.0), principle 5. Umbrella: #112. THE keystone + the research risk.
Goal: prove the epistemic loop on ONE small corpus — regenerate an implementation (or a spec view) FROM the graph, run tests, and feed failures back as "knowledge the implementation had that the graph did not." Give the spec consequences.
Acceptance (spike, not production):
- pick a tiny corpus with tests already in brane's examples/corpus.
- regenerate from the graph; run tests; capture diffs/failures.
- surface each failure as a candidate claim ("production taught this") routed to the promotion gate (#gap2).
- WRITE UP the finding: does the loop round-trip, or does it fail the way MDA/Intentional Programming failed? Verdict in the issue.
This is a spike to de-risk the reframe. Do not over-build; the deliverable is a verdict + a working thread.
Vision: dna/product/vision-spec-machine.md (v4.0), principle 5. Umbrella: #112. THE keystone + the research risk.
Goal: prove the epistemic loop on ONE small corpus — regenerate an implementation (or a spec view) FROM the graph, run tests, and feed failures back as "knowledge the implementation had that the graph did not." Give the spec consequences.
Acceptance (spike, not production):
This is a spike to de-risk the reframe. Do not over-build; the deliverable is a verdict + a working thread.