Connect an AI assistant to Dermal Vectors

1. What you need

• A Dermal Vectors account that belongs to your brand (your company workspace).
• An active 14-day trial or subscription. Check the Billing page.
• An AI app that supports custom connectors (also called MCP servers). This usually means a paid plan of Claude or ChatGPT, or a developer tool such as Codex, Claude Code or Cursor.
• For Option B only: a brand owner to create an API key.

2. Choose a way to connect

See the plan and usage allowances for trial and paid limits, reset dates and how requests are counted.

Both options use the same forecasts and share your brand's monthly allowance. If you're unsure, start with A.

3. Option A: sign in from an AI app

Start with these four steps, then follow the instructions for your app below. Sign-in controls and connector settings vary by app.

1. Copy the server address. Every AI app (ChatGPT, Claude, Codex, Cursor…) uses this same address. It's also on the Integrations page, under “AI apps (sign-in)”:

   YOUR-SERVER-ADDRESS

2. Add it to your AI app as a custom connector or MCP server. App-by-app steps are below.
3. Sign in and approve. Adding the address usually doesn't open anything by itself. Click the app's Connect, Sign in or Authenticate button (each app's steps below say exactly what to click). A Dermal Vectors sign-in page opens in your browser. Check three things:
   • the app name is the one you're using;
   • the brand selected is the right one (use the drop-down if you belong to several);
   • it asks only for organization membership and read released forecasts.
   Then click Allow. If anything looks different, click Deny.
4. Ask a question (examples in section 5). The app appears under “Connected AI apps” on the Integrations page after its first request.

App-by-app steps

Tested in Preview means we verified a connection in our Preview environment. Allowed; not yet tested means the app is allowed to request access, but we haven't verified its connection end to end. These labels describe connection checks, not the accuracy of an assistant's answers.

App menus change often. If a step doesn't match what you see, look for “Connectors”, “MCP servers” or “Integrations” in the app's settings.

codex mcp add dermalvectors --url YOUR-SERVER-ADDRESS
codex mcp login dermalvectors

claude mcp add --transport http dermalvectors YOUR-SERVER-ADDRESS

{
  "mcpServers": {
    "dermalvectors": { "url": "YOUR-SERVER-ADDRESS" }
  }
}

"dermalvectors": {
  "url": "YOUR-SERVER-ADDRESS",
  "headers": { "Authorization": "Bearer ${env:DV_API_KEY}" }
}

4. Option B: use an API key

1. A brand owner creates the key on the Integrations page under “API keys”. Give it a name that says where it's used, and pick an expiry.
2. Copy it straight away. It's shown only once. Save it in a password manager or your tool's secret settings. If it's lost, create a new one.
3. Send it with each request as a header: Authorization: Bearer YOUR_KEY.

With an AI app that accepts a header

Codex, for example, reads the key from an environment variable, so it never appears in the command:

codex mcp add dermalvectors-key --url YOUR-SERVER-ADDRESS --bearer-token-env-var DV_API_KEY

From your own code or automation tool

# List the forecast products this brand can use
curl -H "Authorization: Bearer $DV_API_KEY" https://YOUR-DERMAL-VECTORS-SITE/api/v1/products

# This week's forecast for the metro nearest a location (here: Boston)
curl -H "Authorization: Bearer $DV_API_KEY" \
  "https://YOUR-DERMAL-VECTORS-SITE/api/v1/products/daily_metro_dsi_forecast/latest?near=42.36,-71.06"

Each answer includes an issue ID. To fetch exactly the same data again later, use /issues/ISSUE_ID instead of /latest. Add ?format=csv for a spreadsheet-friendly file.

5. Check your connection and try a question

Check that the tools work

Enable Dermal Vectors in your conversation, then ask:

Use the Dermal Vectors get_product_dictionary tool with product daily_metro_dsi_forecast. Show the DSI units and field definitions returned by the tool.

Expand the tool call in your app to confirm it ran. You should see field definitions and DSI units (a dimensionless index from 0 to 10). This check uses no forecast reads; it confirms the tool connection, but does not confirm access to forecasts.

Next, check forecast access:

Use the Dermal Vectors get_dsi_forecast tool with markets ["14460"], issue "latest", and view "summary". Show Boston's issue ID, forecast dates, mean7 and peak DSI value and date from the returned result.

A successful result identifies Boston, the latest issue ID, the seven-day forecast period, and the returned mean7 and peak DSI values. The dates and values change with each issue; compare the answer with the tool result. If access or current data is unavailable, expect an explicit error instead. An answer without a tool call does not verify the connection. See troubleshooting if either check fails.

The tool catalog lists inputs and results for each tool.

Try a question about your markets

After the checks pass, start with the markets and goals in your existing plan. For example, ask:

Using Dermal Vectors, list the forecast products available to our brand. For the markets in our current plan, show the latest seven-day Dry Skin Index forecast; ask me which markets if you do not know. Cite the issue ID, forecast dates, values, and measure used. Explain higher and lower modeled drying stress as context for questions to test alongside our existing audience and campaign plan. Do not change campaigns or predict results.

For a structured comparison, your brand owner can optionally pin ten population-matched metros. Then ask:

Using Dermal Vectors, call get_pilot_market_comparison for our pinned ten metros. Explain the five matched pairs using the latest seven-day DSI forecast, including where modeled stress is higher and where it is lower. Cite both issue IDs, forecast dates, and values. Keep our existing audience strategy; suggest questions for a marketer to review, not campaign changes or predicted results.

A lower-stress market can be useful context too. It is not, by itself, a reason to stop a campaign or change a budget.

• “Which Dermal Vectors forecast products can I use, and how often are they updated?”
• “Which metro areas in Texas does the Dermal Vectors forecast cover?”
• “Show the 7-day Dry Skin Index for Chicago, and tell me which day is driest.”

A good answer names the forecast issue (an ID like dsi-20261006T140100Z-…) and the dates covered, and quotes the numbers.

One question can trigger several tool calls. Each successful forecast request uses a read from your brand's monthly allowance, even when it returns several markets. Product and coverage lookups use a separate lookup allowance; reading the data dictionary uses no forecast reads. See the plan and usage allowances for details.

6. If something goes wrong

Your AI app usually shows our message in its answer or its log. Find the closest match:

7. Glossary

AI assistant

A chat or coding app powered by AI, such as Claude, ChatGPT, Codex or Cursor.

AI agent

An assistant that can take several steps on its own, such as fetching a forecast, comparing markets and summarizing the result.

MCP (Model Context Protocol)

The standard way AI apps connect to outside services. Dermal Vectors runs an MCP server; its address is what you paste into your app.

Connector

What many apps call an MCP server you've added.

Sign-in / OAuth

The browser window where you approve an app. The app never sees your password.

API key

A secret code your software sends to prove it's allowed in. Treat it like a password.

Bearer header

The standard place to send an API key: Authorization: Bearer YOUR_KEY.

REST API

Plain web addresses that return data, for developers who don't use MCP.

Rotate / revoke / disconnect

Replace a key gradually; stop a key permanently; stop an app (it can be allowed back).

Allowance

The monthly number of forecast reads included with your plan, shared by all your apps and keys.

Brand

Your company workspace in Dermal Vectors. Apps and keys always work for one brand.

Issue ID

The label of one published forecast. The same ID always returns the same numbers.

CBSA

A U.S. Census metro area code, for example 14460 for Boston.

8. Reading the results

The Dry Skin Index (DSI) is a 0–10 forecast of indoor drying stress on skin in a metro area: higher means drier. It comes from temperature and humidity forecasts. It is not search volume, sales, audience size or a medical measure. When comparing markets, quote the issue ID, the forecast dates and the metro areas, so others can check the same numbers.

Other products, such as search-interest outlooks, become available only once they are released for your brand.