Date: May 28, 2026

Importance: High

Status: Needs Follow-up

Trigger: Found wp-content-deployer was mount-only; ran full comparison.

Finding: the canonical repo and live mount have largely diverged

• sail-skills repo (supposed single source of truth): 18 skills
• Cowork /mnt/skills/user mount (what actually runs): 88 skills
• Overlap: only 4 skills
• Mount-only (NOT version-controlled, drift risk): 84 skills
• Repo-only (in repo, NOT loaded in this Cowork session): 14 skills

The repo is not currently functioning as the source of truth. The working core of the library (litify- family, ads/marketing skills, brand-voice-enforcer, pre-publish-qa, sa-design- skills, seo-* skills, etc.) lives only in the mount.

Mount-only skills (84) — exist in Cowork, absent from repo

ad-deploy, ads-budget, ads-create, ads-google, ads-landing, ads-youtube, aguiar-location-page-audit, aguiar-sitemap-page-architecture, ai-seo, ai-visibility-auditor, ai-writing-detection, banana, batch-indexnow, brainstorming, brand-voice-enforcer, campaign-planner, caveman, code-reviewer, content-gap-hunter, copywriting, database-optimizer, dispatching-parallel-agents, fastapi-expert, firm-memory, firm-reference, folder-cleanup, form-refresh, free-tool-strategy, google-ads-daily-negatives, google-ads-mcp, google-ads-optimizer, grill-me, handoff, higgsfield-ops, interactive-element-library, litify-analytics-pipeline, litify-data, litify-flow-builder, litify-intake-ops, litify-integration-ops, litify-knowledge-center, litify-live-recon, litify-marketing-center, litify-negotiation-intelligence, litify-operator-core, litify-report-ops, marketing-analytics, marketing-ideas, marketing-psychology, mcp-developer, notion-session-export, outlook-review, paid-ads, playwright-expert, playwright-visual-qa, pre-publish-qa, programmatic-seo, prompt-engineer, react-expert, react-native-expert, remind-me, reminders, rename-images, sa-editorial-design, sa-postdemand-qa, sa-split-results-hero, sail-design-foundation, sail-design-system, salesforce-developer, schema-markup, screaming-frog-crawl, screaming-frog-ingest, seo-audit, seo-pulse, site-architecture, skill-reconciler, subagent-driven-development, systematic-debugging, test-driven-development, triage-execute-export, using-superpowers, wordpress-pro, wp-site-health, writing-skills. (Plus wp-content-deployer, which was committed to the repo this session.)

Repo-only skills (14) — in repo, not in this Cowork session

banana-router, capacity-watchdog, cloudflare-worker-deploy, cowork-disk-cleanup, github-contents-api-push, github-direct-commit, marketing-competitive-context, monthly-content-refresh, monthly-dedupe, sa-landing-page-layouts, sa-web-layout, seo-competitive-execution, skill-smoke-test, web-ai-concierge.

Why not auto-fixed

A blind bulk push of 84 skills could clobber intentional differences in either direction (some mount skills may be experimental/local; some repo skills may be deliberately retired or newer versions). This needs a reviewed sync plan, not a blind commit. Held for Sam's direction on approach.

Recommended approach (for whoever picks this up)

1. For each mount-only skill, decide: promote to repo (most of them, since they are the working library) or leave local (experimental).
2. For each repo-only skill, decide: is it retired (delete from repo) or should it be loaded into Cowork (sync down)?