test: verify terse review continuation

[khoi]

This temporary PR seeds a clear arithmetic bug, then fixes it in a second commit to verify that SupaReview restores the prior conversation and sends only the terse continuation prompt.

Note

This PR is only for end-to-end validation. It will not merge.

[github-actions]

SupaReview

Findings

The new helper has an off-by-one arithmetic error in its denominator, making its exported result incorrect for normal inputs. The issue is isolated to the added fixture and overlaps the reviewed diff.

Findings

• [P2] Use the total attempt count as the denominator
  

  supareview/fixtures/continuation-review.ts

  Line 2 in 49fe5df

  return successes / (attempts - 1);

  
  successRate subtracts one from the supplied attempt count, so it returns the wrong rate for every normal input (for example, successRate(5, 10) is 5/9 rather than 0.5). It also produces Infinity when there is one attempt instead of the expected 1. The denominator should be attempts itself.

Review type: Full review

Reviewers: thermo-nuclear-review-subagent, thermo-nuclear-code-quality-review-subagent

[github-actions]

SupaReview

LGTM

The continuation patch fixes the prior off-by-one denominator by dividing by attempts and adds an explicit zero-attempt guard. Both review passes found no additional actionable issues in the incremental change.

Review type: Incremental review since 49fe5df...bcb8baa

Reviewers: thermo-nuclear-review-subagent, thermo-nuclear-code-quality-review-subagent

[khoi]

End-to-end validation passed. The restored run received only the terse continuation prompt, derived the prior and current commits from conversation history and Git, reviewed the incremental diff, and confirmed the fix.