automagik-genie
Version:
Self-evolving AI agent orchestration framework with Model Context Protocol support
39 lines (30 loc) • 1.34 kB
Markdown
# QA - Create Collective
## Validation Checklists
### Research Outputs
- Sources cited with links and dates
- Cross-checked key claims (2+ independent sources)
- Bias and scope limits noted
- Evidence snapshots stored under `validation/`
### Writing Outputs
- Audience, purpose, and tone aligned with brief
- Clear structure (title, headings, logical flow)
- Style guide followed (`@.genie/standards/best-practices.md` or project-specific)
- Factual accuracy checks recorded in `validation/`
- Readability pass completed (note decisions in `reports/`)
### Editing & Review
- Line edits: clarity, grammar, concision
- Substantive edits: structure, argument strength
- Before/after diffs or notes saved
- Final approval noted in `reports/`
## Evidence Paths
- Base: `.genie/wishes/<slug>/`
- Validation: `validation/` (drafts, comparisons, citations, notes)
- Reports: `reports/` (approvals, blockers, done reports)
## Critical Edge Cases
- Missing style guide → request or extract inferred guidelines and document
- Ambiguous audience/goal → ask targeted questions; record assumptions
- Source conflicts → note discrepancies and chosen stance with rationale
## Success Criteria
- ✅ Evidence trail demonstrates care and accuracy
- ✅ Tone and structure match the brief
- ✅ Approvals captured before publishing or handoff