# Change log — review pass

The original submission is preserved unchanged in `output/original_submission/` (`handoff.md`, `requirements.csv`). No conclusion changed; the corrections are to sequencing, precision and usability.

## handoff.md

1. **Rollback-plan sequencing corrected.** The original numbered order placed "rollback plan is attached" after Maya's signoff. CR-13 only says it "must be attached first"; it does not order it relative to signoff. The revised order shows it as a precondition of deployment that can be prepared in parallel with the accessibility fix and signoff.
2. **"Four gates" versus five blocked items reconciled.** The bottom line now says four *pre-deployment* gates; section 4 item 5 states that the smoke test and pilot access wait on those four.
3. **Status statements dated.** The accessibility failure is stated as the last recorded result (TEST-11, Oct 26), and signoff as "none recorded as of Oct 28", rather than as unqualified present-tense facts.
4. **Similar project names.** Added that "Northline Atlas" is only the archive title and Atlas Web is the release, alongside the existing Atlas Mobile exclusion.
5. **Pointer to `requirements.csv` added;** minor wording tightened to stay under the word limit.

## requirements.csv

1. **Status column normalised.** Every status now starts with one of eight keywords (At risk, Blocked, Excluded, In force, Not started, Open, Reported met, Staging evidence only) followed by " - " and detail, so the column can be filtered.
2. **Owner inferences removed or labelled.** R-16 (quoted customer request) is now "Not assigned in sources" instead of Maya. R-03, R-04, R-05, R-06 and R-12 label Maya as "CR approver"; R-15 as the person who issued the instruction.
3. **R-07 and R-08 statuses dated** to match the handoff.
4. Row count, IDs and header are unchanged.

## Checks run on the revised files

- Word count of `handoff.md`: 848 by whitespace split, 878 when hyphenated tokens are split, 803 excluding markdown markers; all under 900.
- `requirements.csv` parses with the standard-library CSV reader: required header, 19 data rows, 5 fields per row, no empty cells, unique IDs, ASCII only.
- All 11 record IDs cited in the two files exist in the inputs.
- Sixteen quoted facts (test counts, builds, dates, the 80-account limit, slot time, operator and rollback statements) were string-matched against the input files; all matched.
- All 60 NOTE records were confirmed to share one template; the 6 other records and the register were read in full.
- Input files were not modified (sizes unchanged: 38,697 and 714 bytes).

## Not checked

- No underlying artefacts (test logs, runbook, consent records) are supplied, so the archive's statements about them could not be verified.
- The CSV was not opened in a spreadsheet application; only the Python parser was used.
