Guide Improvement (Interactive Workflow Selector)

Description

Main entry point for guide improvement with interactive workflow selection. Presents user with choice of:

1. Audit EN content only - Fix EN issues, stop before translation for manual review
2. Translate to all locales - Propagate EN content (assumes EN is already clean)
3. Audit + Translate (full workflow) - Both phases with no manual review between

This flexibility allows manual review of EN content before propagating to 17 locales.

Interactive Workflow Selection

When this skill is invoked, it first asks the user which workflow to run before any processing begins.

Why this matters:

• EN content changes affect all 17 locales
• Manual review between audit and translation catches issues early
• Some users want to iterate on EN before translating
• Some users want the full automated workflow

When to Use

Use this skill as the main entry point for all guide improvement work. The interactive selection lets you choose the right workflow for your situation.

Direct sub-skill invocation (skip selection prompt):

• Use lp-guide-audit directly if you always want EN-only processing
• Use guide-translate directly if EN is already clean and you just need translation

When NOT to Use

This skill orchestrates sub-skills. If you need fine-grained control over specific phases or want to skip the selection prompt entirely, invoke the sub-skills directly:

• lp-guide-audit - EN audit and fixes only
• guide-translate - Translation only (requires clean EN)

Inputs

Required

Guide reference (one of):

• Full URL (preferred), e.g. http://localhost:3012/en/experiences/gavitella-beach-guide
• Slug, e.g. gavitella-beach-guide
• guideKey, e.g. gavitellaBeachGuide
• List of URLs/slugs/guideKeys for batch processing

Optional Flags

All flags are passed through to the appropriate sub-skill:

• --skip-json-fix - Skip baseline JSON validation/fixing (passed to lp-guide-audit)
• --skip-validation - Skip EN audit validation in translation phase (NOT RECOMMENDED, passed to guide-translate)

Optional (but recommended)

• Readership/audience (passed to lp-guide-audit)
• Tone constraints + must-include details

Workflow

Step 1: Ask User for Workflow Selection

Use AskUserQuestion tool with:

• Question: "Which workflow do you want to run for this guide?"
• Header: "Workflow Selection"
• Options:
  1. "Audit EN content only" - Run SEO audit and fix EN content issues. Stops before translation so you can review/adjust EN content manually.
  2. "Translate to all locales" - Propagate EN content to all 17 locales (assumes EN is already audit-clean and reviewed).
  3. "Audit + Translate (full workflow)" - Run EN audit/fixes, then immediately translate to all locales (no manual review between phases).

Step 2: Execute Selected Workflow

If user selects "Audit EN content only":

1. Invoke lp-guide-audit skill with guide reference and any flags
2. Report EN audit results
3. STOP - Provide guidance: "EN content is now audit-clean. Review the changes, make any manual adjustments, then run /lp-guide-improve again and select 'Translate to all locales'."

If user selects "Translate to all locales":

1. Invoke guide-translate skill with guide reference and any flags
2. Report translation results for all 17 locales

If user selects "Audit + Translate (full workflow)":

1. Invoke lp-guide-audit skill with guide reference and any flags
2. Gate: Phase 1 must complete successfully before Phase 2 begins
3. If Phase 1 fails: Report failure, STOP, provide guidance
4. Invoke guide-translate skill with guide reference and any flags
5. Report combined results from both phases

Step 3: Report Results

Report results based on selected workflow (see Completion Report section).

Typical Usage Pattern

Most common workflow for careful iteration:

1. Run /lp-guide-improve and select "Audit EN content only"
2. Review the EN content fixes made by the audit
3. Make any manual adjustments to EN content
4. Run /lp-guide-improve again and select "Translate to all locales"

This pattern ensures EN content is reviewed before propagating to 17 locales.

For quick end-to-end processing (no manual review):

• Run /lp-guide-improve and select "Audit + Translate (full workflow)"

Completion Report

If "Audit EN content only" was selected:

Phase 1 Summary (EN Fixes)

• Initial score, final score
• Audit iterations performed
• Key issue categories fixed
• Reader-first improvements made

Next Steps Guidance:

• Review EN changes at the guide URL
• Make any manual adjustments
• Run /lp-guide-improve again and select "Translate to all locales"

If "Translate to all locales" was selected:

Translation Summary

• Locales updated (should be all 17)
• Any locale-specific issues encountered
• Parallel translation efficiency

If "Audit + Translate (full workflow)" was selected:

Phase 1 Summary (EN Fixes)

• Initial score, final score
• Audit iterations performed
• Key issue categories fixed
• Reader-first improvements made

Phase 2 Summary (Translation)

• Locales updated (should be all 17)
• Any locale-specific issues encountered
• Parallel translation efficiency

Overall

• Total guides processed
• Any guides/locales skipped (with reasons)
• Confirmation that:
  • Every write was validated immediately
  • Any corruption was handled only by replacement with known-good content
  • All 17 non-EN locales were successfully updated via parallel subagents
  • No translation work was deferred or suggested for external handling

Core Commitments (Inherited from Sub-Skills)

Every write is validated immediately. After each file write, validate JSON parseability and token integrity before doing anything else.

Readership-first ordering is required. Reorder EN content so the reader can decide/act quickly, with explicit early-content gates.

Localization must not assume sync. Non-EN locale content may have drifted. Check and reconcile entire locale content.

Translation Policy (Non-Negotiable):

• Always complete all translation work in-house
• Always use parallel subagents for localization
• Never defer or skip localization

Error Handling

Phase 1 failure (EN audit):

• Report specific EN audit issues
• STOP before translation
• User can fix manually and re-run, or use lp-guide-audit directly

Phase 2 failure (translation):

• Report which locales failed
• Translation for successful locales is preserved
• User can re-run guide-translate for failed locales

Quality Gates (Combined)

EN (from Phase 1):

1. Final audit score >= 9.0
2. Zero critical issues
3. Zero improvements remaining
4. EN JSON valid + tokens preserved

Translation (from Phase 2): 5. All 17 locale JSONs valid 6. Tokens preserved correctly in all locales 7. Structure parity vs EN in all locales 8. Drift addressed comprehensively

Sub-Skill References

For detailed documentation on each phase:

• Phase 1: See .claude/skills/lp-guide-audit/SKILL.md
• Phase 2: See .claude/skills/guide-translate/SKILL.md