[simurg79]

What is the problem?

Review findings deferred from PR #19 left a redundant placeholder-value test and an overexplained comment.

Fixes #20

How does this PR solve the problem?

Remove the redundant guard and its unused helper, and explain why the two stored values are pinned separately. Exact-value assertions and model figures remain unchanged.

How did you test the PR?

Prior results reported by the user: package tests 89/89 and changed-spec tests 13/13 passed; package type checks and lint passed; formatting passed after an earlier formatting regression was corrected. These checks were not manually rerun during submission.

verified: git push triggered type checks with result: 10 successful, 10 total. git diff --cached --check exited 0 without findings.

Pre-PR Code Review

• Reviewer: Code Reviewer mode (automated)
• Result: ✅ Passed with non-blocking suggestions
• Summary: No blocking findings. One 🟢 nit (local working-copy line endings only; no repo/CI effect).

Agent notes

• Review above is a prior user-reported result: 0 red, 0 yellow, 1 green. The user reported local LF versus sibling CRLF under core.autocrlf=true; repository storage is LF.
• Prior reported build commands: pnpm run check-types and pnpm run lint in packages/types; pnpm exec prettier --check on the scoped test from the root. An earlier prettier --write fixed formatting.
• Prior workflow: spec/design reviewed and acknowledged; no changes needed. Docs, simplification, and post-implementation design review not required. Implementation and commit explicitly approved.
• verified: git diff --stat result: 1 file changed, 1 insertion(+), 15 deletions(-). Scope: vscode-llm.spec.ts.
• verified: git rev-parse HEAD and git ls-remote fork refs/heads/test/issue-20-model-pin-cleanup both returned 2910a29.
• Base: df194f7. Submission only; no source edits or manual test reruns.