← ClaudeAtlas

qa-reportinglisted

Use when preparing QA forecast or outcome Markdown from approved plans and recorded delivery evidence, including DISTILL forecast handoff, blocked delivery summaries, and checking report data against its sources. Not for running tests, deciding quality gates, or publishing prepared Markdown.
SebastienDegodez/skraft-plugin · ★ 8 · Testing & QA · score 68
Install: claude install-skill SebastienDegodez/skraft-plugin
# QA reporting ## Load by task - Before preparing or checking report data, read [report contract](references/report-contract.md) for ownership, fields, source refs and media boundaries. Use its JSON shape; do not add fields. - When referencing gates or evidence, load [quality-gates-evidence-contract](../quality-gates-evidence-contract/SKILL.md) and [skraft-quality-bar](../skraft-quality-bar/SKILL.md). They own evidence semantics and thresholds; this skill defines neither. - For rendering, use the bundled [forecast](assets/templates/forecast.md) or [outcome](assets/templates/outcome.md) template through the existing CLI. Do not copy a template into the consumer project or rebuild its layout by hand. ## Prepare or check data 1. Identify kind, story, full source revision, language and exact returned source refs. Keep generated JSON and Markdown in the consumer's dispatched output directory, never inside the installed skill. Do not reconstruct dated paths. 2. Forecast: project existing approved AC/design and DISTILL plans into criteria, planned tests and sourced expected impact. Do not invent criteria, replan work or describe planned tests as passing. 3. Outcome: reuse the approved forecast/plan and actual evidence, change log and media manifest. State expected versus observed impact, missing proofs and blockers, including when delivery stopped. Delivery is not deployment proof. 4. Preserve source metrics, statuses and limitations. Missing evid