From kai
Core knowledge for using admin-mcp tools: brand codes, URL-to-identifier resolution, identifier format rules, staging-only guardrail, and critical gotchas. Auto-activates whenever admin-mcp MCP tools are being used or when the user asks about page configs, PDPs, widget pages, experiments, or brand settings for Mosaic Wellness brands.
How this skill is triggered — by the user, by Claude, or both
Slash command
/kai:admin-essentialsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are working with the **admin-mcp** MCP server — a staging-only tool for managing Mosaic Wellness page configurations (PDPs, widget pages, experiments, brand settings) stored as JSON in S3.
You are working with the admin-mcp MCP server — a staging-only tool for managing Mosaic Wellness page configurations (PDPs, widget pages, experiments, brand settings) stored as JSON in S3.
All writes via admin-mcp go to staging. Production publishing requires the admin dashboard UI at Zeus (https://stg-zeus.mosaicwellness.in). Never tell users they can publish to production via MCP.
| Hostname Pattern | Brand Code |
|---|---|
*.manmatters.com | mm |
*.manmatters.in | mm-in |
*.manmatters.ae | mm-ae |
*.manmatters.co | mm-co |
*.bebodywise.com | bw |
*.bebodywise.ae | bw-ae |
*.bebodywise.co | bw-co |
*.bebodywise.us | bw-us |
*.ourlittlejoys.com | lj |
*.ourlittlejoys.ae | lj-ae |
*.ourlittlejoys.co | lj-co |
stg-sa.ourlittlejoys.com | lj-sa |
*.littlejoys.us | lj-us |
*.rootlabs.in | rl-in |
*.rootlabs.co | rl-us |
*.onlywhatsneeded.in | wn-in |
*.absolutescience.in | as-in |
Gotchas: lj-sa uses stg-sa. prefix (not stg.). lj-us uses littlejoys.us. rl-us uses rootlabs.co. fw is pagedata-only (no products bucket).
| URL Pattern | Page Type | Identifier | Tools |
|---|---|---|---|
/ | Widget Home | home | get_widget_page_config, update_widget_page_section |
/wlanding/{id} | Widget Landing | {id} | get_widget_page_config, update_widget_page_section |
/product/{urlKey} | PDP (+ experiments) | {urlKey}.json | get_pdp_config, update_pdp_section |
/dp/{urlKey} | PDP (no experiments) | {urlKey}.json | get_pdp_config, update_pdp_section |
/{slug} (catch-all) | Unknown | Try both | Widget page first ({slug}), then PDP ({slug}.json) |
Not supported: /shop/*, /account/*, /consultation/*, /forms/*, /checkout/*, /blog/*
page_name includes .json: minoxidil-5.jsonidentifier does NOT include .json: home, category-hair/^[a-z0-9]+(-[a-z0-9]+)*$/{slug}-exp-{experiment-name}.jsonmm (Man Matters) if not specifiedstaging if not specifieden if not specifiedacquire_page_lock if expiredvariants percentages must sum to exactly 100upload_pdp_image param is file_url, not image_urlpromote_experiment_to_control requires release_notes (mandatory, min 10 chars)add-to-cart, always use the product's url_key (e.g., glp-1-eligibility-test), never the SKU (e.g., abc-3). The storefront checkout page resolves products by URL key. Format: /checkout-v2?add-to-cart={url_key}. Use search_products or get_product_info to find the correct url_key for a product.When a task involves 4+ pages, avoid fetching full configs into context. Use lightweight alternatives:
get_widget_page_summary, get_pdp_summary): Return ~500 bytes per page (widget IDs, types, titles, display order) instead of ~30KB. Use for auditing, comparing structures, or deciding which pages need edits.${CLAUDE_PLUGIN_ROOT}/scripts/bulk-fetch.sh): Downloads full configs to temp files via direct HTTP. Responses never enter context. Use when you need full data for processing.Read ${CLAUDE_PLUGIN_ROOT}/skills/bulk-operations/SKILL.md for the full decision tree and usage patterns.
After ANY write, verify on staging:
preview_widget_page_url or preview_pdp_url)?theme={random_value} to bypass CDN cachetheme value if changes don't appear (up to 2 retries)For deeper knowledge, read files from ${CLAUDE_PLUGIN_ROOT}/references/admin/:
workflow-patterns.md — Step-by-step multi-tool workflowsroute-reference.md — Complete URL resolution algorithmbrand-bucket-reference.md — 18 brands with S3 buckets and preview URLswidget-type-catalog.md — 122 widget types + 41 action typesdisplay-order-guide.md — 3-tier displayOrder override systemexperiment-lifecycle.md — Full A/B test lifecyclepdp-transformation-pipeline.md — 7-step middleware enrichmenttool-reference.md — All 38 tools with "when to use" guidanceerror-handling.md — Common errors and resolutionsnpx claudepluginhub mosaic-wellness/ai-toolkit --plugin kaiGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.