research-document
Import Decisions
Import Decisions
D-001: Use the migration-plan content model for subject matter
- Issue: The repository already contained
knowledge-platform/Migration-Plan.mdwith explicit destination paths for the handbook and course materials, while the bootstrap kit proposed a different root structure. - Evidence: The migration plan is repository-local, current, and specifically maps the imported handbook and course files into
content/. - Decision: Use the migration-plan destinations for handbook and course artifacts.
- Consequences: The imported learning content now lives under
content/, while bootstrap governance and operational docs were integrated elsewhere. - Reversibility: High. Paths can be moved again later if a stronger canonical model emerges.
D-002: Keep the existing REP v2 file as canonical
- Issue: The bootstrap kit included
specifications/research/Research-Execution-Package-Specification-v2.md, but the repository already hadai-prompts/Research-Execution-Package-Specification-v2.md. - Evidence: The files share the same SHA-256 hash,
eff69198cdce0e874d8fa66195d81be96aa122362f08ac1ead6469c82b0a82b0. - Decision: Keep the existing
ai-prompts/Research-Execution-Package-Specification-v2.mdas the sole active canonical REP v2 file and remove the duplicate intake copy. - Consequences: No duplicate active REP v2 artifacts remain.
- Reversibility: High.
D-003: Promote bootstrap repository controls into active root/docs locations
- Issue: The repository lacked root entry-point documents such as
README.md,CURRENT-STATE.md,ROADMAP.md, andCHANGELOG.md. - Evidence: The bootstrap kit supplied these artifacts, and no stronger competing canonical versions existed in the repository.
- Decision: Move the bootstrap entry-point documents into active repository locations, then patch references to the actual canonical paths.
- Consequences: The repository now has live project controls and agent instructions.
- Reversibility: High.
D-004: Preserve bootstrap placeholders as historical scope markers
- Issue: The bootstrap kit included placeholder README files for CI workflows, tests, tools, and website generation.
- Evidence: The files are minimal and do not represent active implementations, but they do capture intended scope.
- Decision: Preserve them under
docs/repository/with bootstrap-placeholder names instead of treating them as active directories. - Consequences: Scope intent is retained without implying implemented tooling.
- Reversibility: High.
D-005: Preserve the Chapter 1 ZIP as derived historical output
- Issue:
Chapter-01-Why-AI-Engineering-Is-Different.zipis a packaged derivative rather than canonical source material. - Evidence: The ZIP contents are chapter deliverables derived from the same lesson and research material already imported as Markdown.
- Decision: Archive the ZIP under
archive/packages/ai-engineering-course/. - Consequences: Provenance is preserved without elevating the archive to canonical source status.
- Reversibility: High.
D-006: Preserve the lesson draft as a draft, not a second canonical lesson
- Issue:
Lesson 2.mdis internally titledLesson 1 (Working Draft). - Evidence: Filename and internal title conflict; content is too short to establish an authoritative second lesson.
- Decision: Preserve it as
content/projects/ai-engineering-course/chapter-01/drafts/lesson-01-working-draft.md. - Consequences: The draft remains available without competing with the canonical lesson.
- Reversibility: High.
D-007: Recreate the empty research directories outside the intake area
- Issue: The bootstrap kit shipped only
.gitkeepplaceholders for research working directories. - Evidence: The directories communicate the intended operating model, but leaving them under
input-documents/would violate the cleanup requirement. - Decision: Recreate the placeholder directories under
research/and remove the intake copies. - Consequences: The canonical repository model now includes empty working locations without retaining a nested bootstrap tree.
- Reversibility: High.