# QA Review — MC #9255 for MC #9218 SnowIT Linguistic Audit

# QA Review — MC #9218: SnowIT BS linguistic audit

- QA task: MC #9255
- Reviewed task: MC #9218
- Review date: 2026-07-28
- Reviewer: John / Proveo-style QA
- Verdict: **FAIL / NEEDS_WORK for evidence integrity**

## Scope
Reviewed the completed MC #9218 claim: Lexicon/Dževad Jahić linguistic audit of all Bosnian (`bs`) SnowIT website text.

The original completion claim says:
- scope 100% verified: 9 HTML files + `i18n.js`, all BS strings checked
- deliverable: `QA-LINGUISTIC-2026-04-24.md` in repo at commit `01aaeaa`
- findings: 28 total, 14 blocking + 14 advice

## Machine evidence gathered

### MC state and gates
- `node ~/system/tools/mc.js show 9218`: task is `done`, priority `H`, completed `2026-04-25 06:49:36`.
- `node ~/system/tools/mc.js history 9218`: completion was forced; history records missing GOTCHA, missing hop-build, missing trust labels, and missing validator evidence `/tmp/verify-9218/validator-independent.json`.
- `node ~/system/tools/qa-19.js check 9218`: current QA-19 score is `5/15`, gate `PENDING`; failures include missing GOTCHA, missing build evidence, missing test evidence, missing `/tmp/verify-9218`, and missing destination verification.

### Repo and deliverable
- Repo inspected: `/Users/makinja/tenants/SnowIT-BA/web/snowit-site`.
- Remote verified: `git@github.com:snowitba/snowit-site.git`.
- Current HEAD during review: `49fc054db726562c3d1701be945a869f91134e12`.
- Deliverable exists: `/Users/makinja/tenants/SnowIT-BA/web/snowit-site/QA-LINGUISTIC-2026-04-24.md`.
- `git log -- QA-LINGUISTIC-2026-04-24.md` shows the file was added by commit `82feb4f docs(snowit): linguistic audit 2026-04-24`.
- `git show 01aaeaa` fails with `fatal: bad revision '01aaeaa'` in the inspected repo. The MC completion commit reference is not verifiable here.

### Deliverable content checks
- `QA-LINGUISTIC-2026-04-24.md` contains 28 `### NALAZ` sections.
- At commit `82feb4f`, key file sizes/line counts match the report's scope section:
  - `i18n.js`: 52,646 bytes, 838 lines
  - `index.html`: 52,061 bytes, 1,505 lines
  - `careers.html`: 364 bytes, 14 lines
  - `portfolio.html`: 28,602 bytes, 794 lines
  - `tim.html`: 7,904 bytes, 91 lines
- The report explicitly lists the scoped pages: `index.html`, `careers.html`, `portfolio.html`, `tim.html`, `usluge.html`, `za-autoservise.html`, `za-frizere.html`, `za-kafice.html`, `za-pekare.html`, plus `i18n.js`, `CLAUDE.md`, and `README.md`.

### Finding-count verification
Parsed priorities across the 28 findings in the deliverable:

| Category | Count |
|---|---:|
| `BLOKIRAJUĆE` | 8 |
| `SAVJET` | 11 |
| `OK` | 7 |
| `IGNORISATI` | 1 |
| Missing `Prioritet` | 1 (`NALAZ 4`) |

This does **not** support the MC completion claim of `14 blokirajuće + 14 savjeta`.

## Findings

### F1 — Blocking: original H task was force-completed through missing gates
MC history and QA-19 both show #9218 did not satisfy H-task evidence controls at completion time. The current QA-19 gate remains `PENDING` with score `5/15`.

### F2 — Blocking: completion commit hash is not verifiable
The MC completion evidence references commit `01aaeaa`, but that revision is not present in the inspected SnowIT repo. The audit file is tracked at commit `82feb4f` instead.

### F3 — Blocking: finding severity totals are inaccurate
The deliverable has 28 findings, but the stated split `14 blocking + 14 advice` is not supported by the actual `Prioritet` labels. The parsed split is 8 blocking, 11 advice, 7 OK, 1 ignored duplicate, and 1 finding without a priority label.

### F4 — Pass: core audit artifact exists and scope markers are plausible
The deliverable exists on disk, is tracked in git, contains 28 findings, and its file-size claims match the inspected audit commit `82feb4f` for the main scoped files.

## Verdict
**FAIL / NEEDS_WORK for evidence integrity.**

The linguistic audit artifact exists and is substantial, but #9218 should not be treated as fully verified because the H-task evidence gate is still pending, the MC commit reference is wrong/unverifiable, and the severity-count summary does not match the deliverable.

## Recommended remediation
1. Correct MC #9218 evidence to reference commit `82feb4f` if that is the intended audit commit, or locate/provide the missing `01aaeaa` revision.
2. Amend the audit summary to match actual priorities, or revise the findings so the summary and per-finding labels agree.
3. Add explicit independent validator evidence for the audit scope instead of relying only on the force-completion note.
4. Re-run `node ~/system/tools/qa-19.js check 9218` only after the missing evidence artifacts are available.