Reports¶
Human-facing reporting follows ADR-0008: every artefact answers exactly one named question from the question catalog, and a report is an assembly of such units. Three layers live here:
| Layer | What | Where |
|---|---|---|
| Question units (human) | One generated Markdown file per question — story progress, scoreboard, test traceability, ADR links, code views | units/ |
| Audit products (machine/audit) | Stable CSV matrices per ADR-0006 layer 3 | trace/ |
| Live answers | The commands themselves — always current, never stale | arqix trace … |
Units¶
- How far along is each user story? (Q-01)
- What share of the requirements is verifiably implemented? (Q-03)
- Which tests verify which requirements? (Q-02)
- Which user story belongs to which integration test? (Q-05)
- Which workflow belongs to which integration test? (Q-06)
- Which ADRs are linked to which requirements? (Q-07)
- Which code implements which requirement? (Q-04)
- Where is the documentation for a given piece of code? (Q-08)
- How large is the codebase? (Q-09)
- What is the code coverage of the test suite? (Q-10)
Regeneration and staleness¶
All committed files here are snapshots (each carries its commit + date in a generated header) and go stale with every change to requirements, tests, or markers. Refresh with:
arqix report snapshot --stamp "<sha>, <date>"
arqix trace matrix > docs/en/reports/trace/matrix.csv
arqix trace matrix --type us-req > docs/en/reports/trace/matrix-us-req.csv
Regeneration stays manual (just reports), but staleness is gated: arqix report snapshot --check runs inside arqix verify (and therefore in CI) and fails when any committed snapshot no longer matches the corpus.
The live answer is always the command.