docs: add handoff readiness scale

docs/review/HANDOFF_READINESS_SCALE.md

@@ -0,0 +1,75 @@
+# Handoff Readiness Scale
+
+OpenAMP Foundry should distinguish exploratory dry-lab artifacts from artifacts that are ready for qualified external review.
+
+This scale defines evidence readiness for handoff packets without adding wet-lab protocols, operational methods, candidate-generation instructions, or optimization objectives.
+
+## Readiness levels
+
+| Level | Name | Meaning |
+|---|---|---|
+| H0 | exploratory | Useful for internal learning only. Not ready for external review. |
+| H1 | structured | Artifact is valid, reproducible, and documented, but major evidence gaps remain. |
+| H2 | reviewable | Artifact has enough provenance, baseline comparison, limitations, and safety boundaries for expert review. |
+| H3 | bounded review | Expert review may proceed only inside a clearly narrowed scope. |
+| H4 | handoff ready | Artifact is suitable for qualified external review under the documented claim boundary. |
+| H5 | returned evidence | External review or result has returned and must update the evidence record. |
+
+No level means biological efficacy is proven.
+
+## Minimum handoff readiness note
+
+```markdown
+## Handoff readiness
+
+**Artifact:** Evidence certificate, shortlist, report, or review packet.
+
+**Readiness level:** H0 | H1 | H2 | H3 | H4 | H5.
+
+**Reason:** One paragraph.
+
+**Required evidence present:** Provenance, run manifest, baseline comparison, leakage audit, limitations, safety boundary, reviewer route.
+
+**Missing evidence:** What prevents the next level?
+
+**Allowed use:** Internal learning | maintainer review | expert review | bounded external review | update from returned evidence.
+
+**Not allowed:** What this artifact must not be used to claim.
+```
+
+## Promotion guidance
+
+Promote readiness only when the artifact becomes safer and more reviewable, not merely more impressive.
+
+An H4 artifact must have:
+
+- clear provenance;
+- reproducible generation path;
+- declared claim boundary;
+- baseline comparison when model-like components are involved;
+- leakage audit when evaluation is involved;
+- limitations and failure modes;
+- reviewer routing;
+- explicit statement of what is not claimed.
+
+## Reviewer guidance
+
+Reviewers should use the scale to avoid binary thinking.
+
+A packet may be good work and still only H1 or H2. That is not failure. It is honest status.
+
+## Agent guidance
+
+Agents should propose a readiness level before requesting external review.
+
+If an agent cannot say what evidence is missing for the next level, the artifact is not ready to leave internal review.
+
+## Acceptance test
+
+The readiness label is useful when a maintainer can answer:
+
+1. Who may review this artifact?
+2. What evidence is present?
+3. What evidence is missing?
+4. What claim boundary travels with it?
+5. What must not be claimed from it?