**Overall Verdict:** Strong Accept — The submission is complete, reachable, and supported by exact reference certificates whose substantive mathematical checks were independently confirmed.

**Completeness Check:**

- Instruction: Complete — Clearly defines the optimization problem, degree bounds, exactness requirements, and required deliverables.
- Inputs: Complete — Solver-visible task and certificate-format files are sufficient and separated from reference materials.
- Toolchain: Complete — The open-source Python workflow is suitable for a single machine and requires no external service.
- Reference output: Complete — Includes explicit primal and dual JSON certificates, derivation, verification source, configuration, and verification log.
- Rubric: Complete — The 10/45/45 allocation objectively distinguishes exact identification, primal attainment, and complete dual certification.

**File and Task Reachability:**

Attachments: `reachable`. Task: `reachable`.

Inspected `optimal_sos_n12_inputs.zip`, `optimal_sos_n12_reference.zip`, and `optimal_sos_n12_evidence.zip`, including the task, certificate format, both reference certificates, derivation, verification source, and trial evidence. All three archives matched their manifest byte counts, passed archive-integrity checks, and contained readable UTF-8 text or JSON without unsafe paths or encryption. All 43 members were inventoried; large exploratory numerical records received structural and representative-content inspection.

Independent exact checks confirmed the reference optimum, every ordinary primal coefficient identity, basis substitutions, degree bounds, Gram symmetry and factorizations, and positivity of all 21 primal blocks. Dual checks confirmed normalization, objective equality, all seven positivity blocks, and reconstruction of all 140 moment-matrix entry orbits. The finite moment rule and parity reduction cover the complete ordinary polynomial spaces and make every required localizer zero. Uploaded programs were inspected, not executed.

The completion path is coherent: derive certificates from the solver-visible inputs, submit exact JSON data with a derivation and verifier, and evaluate the three rubric criteria. The self-test documents one incomplete, zero-score trial; it does not establish a population success probability.

License keys, accounts, tokens, and other credentials are not required from the contributor. Confidential, proprietary, or internal material may be evaluated as submitted evidence without requiring additional disclosure.

**Special Screening:**

None. Trial evidence cites related published research, but public availability of this exact task or protected solution was not independently established.

**Recommended Changes:**

The complete exact reference and transparent reporting of the unsuccessful self-test support readiness. No changes are required.