fix(rubric): collapse platform factors into single 2-point factor

[stack72]

Summary

• Mirrors swamp-club#509 — collapse platforms-one and platforms-two into a single platforms factor worth 2 points
• A single-platform manifest now earns full credit — platform breadth is not a quality proxy when the upstream constrains it
• Factor count drops from 10 to 9; max stays at 12
• CLI and registry now produce identical scoring

Test plan

• All 66 unit tests pass (deno test src/domain/extensions/extension_rubric_scorer_test.ts)
• Parity tests updated to match swamp-club server scorer
• Integration test updated (single-platform manifest no longer fails platforms-two)
• Run integration/extension_quality_test.ts in a full dev environment

🤖 Generated with Claude Code

[github-actions]

Adversarial Review

Critical / High

None.

Medium

1. src/domain/extensions/extension_rubric_scorer.ts:351 — tautological condition

   manifest.platforms.length === 0 || manifest.platforms.length >= 1,

   This is always true for any array — .length is either 0 (first branch) or ≥1 (second branch). There is no possible input that causes the platforms factor to be "missing".

   This appears to be intentional per the PR description ("single-platform manifest now earns full credit"), and the old platforms-one factor had the same tautology, so behavior is preserved. But the condition is misleading — a reader will wonder what input could fail it. Consider simplifying to just true to make the intent explicit, or adding a brief comment that the factor is always earned when platforms is a valid array (i.e., the factor gates on the field existing, not its value).

   Not blocking because behavior is correct and intentional.

Low

None.

Verdict

PASS — Straightforward collapse of two factors into one. Arithmetic is correct (9 factors, 12 max points preserved). Tests are updated consistently across unit and integration layers. The only concern is a cosmetic tautology in the boolean condition.

[github-actions]

Code Review

Blocking Issues

None.

Suggestions

1. Tautological condition (extension_rubric_scorer.ts:351): The guard manifest.platforms.length === 0 || manifest.platforms.length >= 1 is always true — every array satisfies one of those branches. This was inherited from the old platforms-one factor, so it's not a regression, but it could be simplified to just true to make the intent ("always award these points") self-evident. Minor readability nit, not blocking.

---

DDD: Changes are well-contained in the domain layer (src/domain/extensions/). composeScore is pure domain logic operating on value objects — no boundary violations.

Tests: Thorough — 66 unit tests updated, parity tests aligned, integration test updated. Factor count (9) and max points (12) are verified.

CLAUDE.md compliance: License header present, tests adjacent to source, no internal libswamp imports from outside the boundary, comment in the doc block updated to reflect the new factor count.

Clean, well-scoped change.