Background
PR #120 added .gitattributes.template and a first-sync bootstrap so consumer repos get a starter .gitattributes (consumer-owned thereafter). However, .github/copilot-instructions.md says nothing about line endings, so future agents working in a consumer repo may still default to dos2unix or other ad-hoc fixes when a user reports CRLF issues. That is the exact problem PR #120 was meant to head off.
Proposed change
Add a short note to .github/copilot-instructions.md (3-5 lines), either under "Tool Preferences" or as a new "Line Endings" subsection, covering:
- Line endings are governed by
.gitattributes.
- Never run
dos2unix or other bulk line-ending conversion tools.
- Policy edits go in the consumer's own
.gitattributes, which is consumer_owned -- Pull-SDLC.ai.ps1 never overwrites it after first sync.
Acceptance
- New text in
.github/copilot-instructions.md covers the three bullets above.
- Validate Instructions workflow stays green.
Out of scope
Context
Spawned from PR #120 follow-up review.
Background
PR #120 added
.gitattributes.templateand a first-sync bootstrap so consumer repos get a starter.gitattributes(consumer-owned thereafter). However,.github/copilot-instructions.mdsays nothing about line endings, so future agents working in a consumer repo may still default todos2unixor other ad-hoc fixes when a user reports CRLF issues. That is the exact problem PR #120 was meant to head off.Proposed change
Add a short note to
.github/copilot-instructions.md(3-5 lines), either under "Tool Preferences" or as a new "Line Endings" subsection, covering:.gitattributes.dos2unixor other bulk line-ending conversion tools..gitattributes, which isconsumer_owned--Pull-SDLC.ai.ps1never overwrites it after first sync.Acceptance
.github/copilot-instructions.mdcovers the three bullets above.Out of scope
.gitattributes.templatecontent..gitattributesto be union-merged like.gitignore(explicitly rejected during PR feat(sync): upstream-only gitignore markers + .gitattributes.template bootstrap #120 discussion).Context
Spawned from PR #120 follow-up review.