The `dintel-gtm-toolkit` Python scripts (analyze_container.py,
diff_versions.py, validate_tags.py, find_unused.py) have been
replaced by the DTM Agent skill set, which operates on live
containers via the GTM API instead of exported JSON.
- .claude/commands/gtm-guardian.md: rewrite as a deprecation
pointer with old→new mapping for every script/phase
(Phase 6 → /dtm-audit + /dtm-debug, Phase 7 → /dtm-taxonomy +
/dtm-lookup, etc.).
- custom-skills/62-gtm-validator/code/references/phase6-audit.md:
replace "D.intelligence GTM Toolkit Integration" section with
DTM Agent skill map, live-container workflow, and dtm CLI
examples; update objective #4.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
gtm-audit (60):
- Added Mode D rule: check elements for IDs before designing CSS triggers
gtm-editor (61):
- Added "Tagging Workflow: dataLayer First" section — always suggest
dataLayer push code for developers before creating GTM triggers
- Added multi-language snippets (JS, React, Vue, PHP)
- Added "Trigger Design: IDs First" section — prefer element IDs,
ask user to add IDs before falling back to CSS selectors
- Added ID naming convention: [section]-[element]-[action]
- Added decision tree for trigger selection
gtm-validator (62):
- Flag CSS-based triggers as fragility risk during QA
- Recommend adding IDs instead of finding alternative CSS selectors
Based on field testing with ourdigital.org GTM-N9TPJW container.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
