Promote dcr-v1-to-v2 from Snowflake-Solutions

[jdanielmyers]

Summary

Promotes the dcr-v1-to-v2 skill from the internal Snowflake-Solutions/cortex-code-skills repo to public Labs, rewritten for an external builder audience (developers, data engineers).

Run through the v1.1.0 audit pipeline:

• Holistic LLM rewrite for builder audience
• All internal Snowflake references substituted with public equivalents (e.g., Snowhouse -> ACCOUNT_USAGE)
• audience.no_internal_refs check passes

Test plan

• Audit blocking_failures == 0
• audience.no_internal_refs passes
• Frontmatter matches Labs conventions (snowflake-docs / manage-zerocopy-sapbdc precedents)
• LICENSE present (Snowflake)
• No Snowflake-Solutions repo path leakage

[jdanielmyers]

Audit Report: dcr-v1-to-v2

Re-audit (post-v1.1.1 + critical fixes). Original report superseded.

Summary

Category	🔴	🟡	🟢
Frontmatter	0	0	0
Structure	0	1	0
Workflow	0	0	1

Critical 🔴

None — all critical findings from the original audit have been addressed.

Warnings 🟡

1. Sub-flow generate/INSTRUCTIONS.md is 739 lines — significantly over the 500-line guideline. This sub-flow handles Collaboration API script generation and contains extensive SQL templates. → Consider splitting the SQL template blocks into a references/templates/ directory and keeping generate/INSTRUCTIONS.md as the workflow orchestrator.

Suggestions 🟢

1. The router SKILL.md is well-structured with clear 6-step workflow, 4 stopping points, comprehensive "Common Mistakes" section (especially the two-call async teardown pattern and | join_policy filter removal).

Verdict

Ready to merge — Router is clean and well-structured. The 739-line sub-flow is a non-blocking concern (content is valid, just long). Merge and consider refactoring the sub-flow in a follow-up.