GlowTheory
Connect an agent to GlowTheory
GlowTheory runs an MCP server, so an assistant can work with your actual skin profile, routine and shelf instead of giving generic advice. Connecting takes one URL and a sign-in.
The URL
https://mcp.glowtheory.app/mcp
Paste that into any MCP client. You will be sent to a GlowTheory sign-in page, and that is the whole setup. There is no API key to generate and no application to register.
If a connector asks for a Client ID, leave it blank. The server supports dynamic client registration, so the client issues itself an ID. That field exists for servers that cannot do this.
Meta Muse is the exception. Its form will not let you continue with the field empty. Enter:
glow
That ID only works from Meta's connector — it is tied to Meta's callback address, so it cannot be used to route your sign-in anywhere else. In every other client, leave the field blank.
Muse may then show “something went wrong” on the add screen even though the connector was added. Close the dialog and start chatting — asking it about your routine or your shelf will confirm it is working.
Claude Code
claude mcp add --transport http glowtheory https://mcp.glowtheory.app/mcp
What it can see
- profile:read
- Skin type, concerns, experience level, philosophy
- routine:read
- Current AM/PM routine steps and ingredients
- shelf:read
- Products owned and shopping list
- shelf:write
- Save products, skin check-ins and event prep plans to your account
Every tool acts as the signed-in user and reaches only that user's data. You can disconnect from your client at any time.
Tools
Read your context
- get_my_glow_context
- Profile, concerns, routine, shelf and shopping list in one call. Start here.
- get_my_profile
- Skin type, concerns, experience level, skincare philosophy.
- get_my_concerns
- Current concerns plus the latest skin-scan score.
- get_my_routine
- Ordered AM/PM steps with product type and key ingredients.
- get_my_shelf
- Products owned, and the shopping list.
- get_my_ingredient_overview
- Repeated ingredients and categories across routine and shelf.
Analyse
- analyze_my_routine_safety
- Active conflicts, irritation load, pregnancy warnings, scheduling cautions.
- explain_ingredient_for_me
- What an ingredient does, against this user’s concerns and routine.
- analyze_product_for_me
- Whether a product fits this user, from details or OCR’d ingredients.
- recommend_treatments_for_me
- Ranks professional treatment options. Educational, not medical advice.
Save back to your account
- add_to_shelf
- Save a product to the user’s shelf.
- add_to_shopping_list
- Save a product or idea to the shopping list.
- log_skin_checkin
- Record how the user’s skin feels today.
- create_event_prep_plan
- Build and save a plan for a wedding, shoot or other event.
Catalog and reference
- search_products
- Search the GlowTheory product catalog.
- get_reference_clinics
- Reference clinic data.
For implementers
- Transport: streamable HTTP
- Auth: OAuth 2.1, dynamic client registration (RFC 7591), PKCE with S256
- Discovery:
https://glowtheory.app/.well-known/mcp.json - Resource metadata:
https://mcp.glowtheory.app/.well-known/oauth-protected-resource - Agent-readable summary:
https://glowtheory.app/llms.txt
Scope and limits
GlowTheory is cosmetic, not medical. Skin scans produce aesthetic observations on a 0–10 scale, not diagnoses, and treatment recommendations are educational. Routine logic is hand-coded from accredited skincare curricula, including ingredient conflict rules, frequency guidelines and pregnancy-safe swaps. Full detail on data handling is in our security and privacy notes.
New to GlowTheory? Start here. Questions: hello@glowtheory.app.