# Data audit (P1)
| file | SHA-256 | lines (wc -l) | matches README/bundle hash |
|---|---|---|---|
| data/README.md | fc449910f4cab76a205f793c65e408d852da63a3ae11b49a90c71dc3cf27f13c | 9 | yes (bundle.json) |
| data/sources/statcan-daily-lfs-2026-09.txt | 7ea5fec57c87874ff5ee09a8304927a1aaed5a1837db5312fab9f7a2d5f2d0bb | 164 | yes (README and bundle.json) |

Schema: plain text; 10-line provenance header, then release text with `## ` section headings (17 `##`-level headings incl. Chart/Infographic/Map placeholders) and `###` note sections. Units: percent for rates, persons for employment changes; seasonally adjusted unless stated. Reference period September 2026 (reference week Sept 13–19, 2026); released 2026-10-09; retrieved 2026-10-30.
Missing: data tables and charts (headings only) — not needed. Duplicates: headline figures stated twice (Highlights and dedicated sections) and agree (6.5%, -68,000).
Anomalies: README contains template text about RESEA/London tables and regression tables unrelated to this extract; ignored.
Information-availability: all input is the admissible extract; no violating input. No external values used.
No rows filtered (text extraction task); exclusion ledger records a no-op step.