A due-date fix, reviewed in a draft pull request

What happened

A new private synthetic repository held four baseline files. We created a correction branch, changed the overdue boundary from <= to < and opened an unassigned draft pull request. The saved diff contained one changed file, two additions and two deletions, including the comment correction. Main retained the seeded bug. Local Node tests run from the rendered GitHub source readback changed from two passes and one expected failure to three passes. GitHub Checks showed zero checks; no Actions or Copilot was used. The browser file picker stalled, so we completed the commits through the web text editor. The PR remained draft and unmerged.

Assertions and observed results

• PassPrivate baseline

  Expected: A new private synthetic repository holds README, package metadata, implementation and three tests.

  Observed: GitHub showed Private and exactly the four named root files after four baseline commits. No workflow folder was added.

• PassBaseline failure

  Expected: The seeded <= comparison fails only the due-today case.

  Observed: Local Node execution of the fixture, then the rendered GitHub source readback, both returned 2 pass / 1 fail: due today returned true instead of false.

• PassBranch isolation

  Expected: The correction is on a separate branch and main retains the seeded bug.

  Observed: GitHub showed two branches, the correction branch one commit ahead; its file contains < while the main file view still contains <=.

• PassCorrected tests

  Expected: Past-due open=true, due-today open=false, paid=false.

  Observed: Local Node execution using the saved GitHub branch file-view readback returned all three tests passing, exit 0. These tests ran locally, not in GitHub Actions.

• PassDraft pr

  Expected: Create an unassigned draft PR with no reviewers and inspect its actual diff.

  Observed: GitHub confirmed Draft, no reviews, no assignees, one commit, one changed file, two additions and two deletions. Diff contains the boundary comparison and comment correction. Checks(0); Merge pull request disabled.

• BlockedUpload transport

  Expected: Upload the four synthetic fixture files through the browser file picker.

  Observed: The file-chooser automation stalled and was aborted; a fresh upload view contained no files. This transport was not validated. All four files were subsequently entered and committed through the normal web text editor.

Reproduce the task

1. Run node --test invoice.test.js on the seeded fixture and save the expected one-test failure.
2. Create a new private scratch repository with no collaborators, workflows or real data.
3. After the browser upload transport stalls, cancel it and commit the four small fixture files through GitHub’s text editor.
4. Create a correction branch from the full baseline; change <= to < and remove the intentional-bug wording from its comment.
5. Read branch and main implementation, package metadata and tests back from the file views; run the readback locally.
6. Create an unassigned draft PR; inspect the saved PR Files changed view. Leave the PR draft and unmerged.

Inputs and saved outputs

The downloads contain synthetic test material. Account identifiers and private project links are excluded. Hashes identify the exact published bytes.

• Code-only screenshot of the observed GitHub draft diff ↓SHA256 e4c8d1769408fdfe15ce2dfb871804929c0b08f199ed62ac735928ab2bd7f6b2
• Synthetic source diff reconstructed from rendered branch/main readback ↓SHA256 83ad29630eac64e4b716e36955049b7b2fca5f9d2d015118d387eb95734267da
• Original fixture Node test failure ↓SHA256 97dd514d4ce2be3e9b859a713b19b406b970dc4fda67e49359435fb1b5d0514d
• Corrected local fixture Node test pass ↓SHA256 3de3efb314bfb90de52f17ed6cab714a418b2f9d9c554af87c72491ed426e2ec
• Rendered GitHub main source Node test failure ↓SHA256 ac73baf06c11167000a9cb1e1625abbd08a87cb283149f960ce53ed4eaba8dfc
• Rendered GitHub branch source Node test pass ↓SHA256 a48d46eb805482be46a67d927769c7d8f8eb046cab824f39020548fb8071e759
• Executable synthetic main/branch source, diff and local test outputs ↓SHA256 6ca4165d8536461922983f0337e24da73ee4498afcbd76d0d846430667b0309a

Limits of this result

• This is one synthetic workflow, not evidence of general performance, reliability, security or usability.
• No Copilot, AI code quality, Actions, CI, collaboration, reviewer notification, merge or paid-plan feature was tested.
• The browser upload transport failed; the web text editor was the tested successful path.
• The rendered source readback omits a trailing newline, so its checksum is distinct from a raw-file export.
• Repository identifiers and account names remain in private receipts. The public screenshot is captured from only the synthetic diff region.
• GitHub had one recorded exact profile pageview in the available 30-day Vercel sample; owner/test visits cannot be excluded. This does not establish popularity.

This record does not change a general product score or name an overall winner. Changes to editorial opinions or recommendations require review.

Continue the research