DataScaler provides a Remote MCP service for Claude and Claude Code. After authorization, Claude can query brand data available to the user's DataScaler account, including mentions, sentiment and topic changes, platform distribution, competitor comparisons, and source posts.
DataScaler remains the system for collection, brand configuration, permissions, and source records. Claude receives the question, calls the MCP tools, and formats the response. The connection is intended for teams that want to analyze DataScaler data from Claude without replacing the DataScaler dashboard or report workflow.
What DataScaler MCP provides
DataScaler MCP returns data for brands the current account is authorized to access.
Capability | Returned data | Typical use |
|---|---|---|
Brand performance | Mentions, engagement, sentiment, and topics for a specified period | Weekly reviews, monthly reports, and launch analysis |
Platform analysis | How public platforms contributed to changes in volume, engagement, and topics | Identifying where a change occurred |
Competitor comparison | Performance of configured competitors and contexts in which brands are discussed together | Competitor tracking and purchase-context analysis |
Customer feedback | Positive and negative themes, recurring questions, and representative discussions | Routing findings to product, support, and content teams |
Source verification | Public posts and source links supporting a result | Checking context before using a summary |
The MCP service returns data collected and organized by DataScaler. It does not ask Claude to crawl the open web. Coverage depends on brand configuration, account permissions, supported platforms, the requested time range, and collection status.
How a query is handled
A typical request follows four steps:
- The user asks a question with a brand, time range, and analysis goal.
- Claude selects a DataScaler MCP tool and sends the query parameters.
- DataScaler checks OAuth, organization access, and brand permissions, then returns data and sources.
- Claude formats the result. The user can continue with narrower questions about a platform, competitor, theme, or source post.
DataScaler does not read unrelated Claude conversations. The MCP service also does not provide actions for publishing, commenting, or replying on social platforms.
Connect DataScaler Remote MCP
DataScaler currently uses one Remote MCP endpoint:
https://app.datascaler.ai/mcp
Claude on the web or desktop
Open Customize → Connectors in Claude:
- Click + Add → Add custom connector.
- Enter
DataScaleras the name. - Enter
https://app.datascaler.ai/mcpas the Remote MCP URL. - Under Authentication, choose Sign in now or Sign in when needed.
- Under OAuth client, choose Register automatically.
- Click Connect, sign in to DataScaler, and review the account, organization, and requested access.
- In a conversation, use + → Connectors to enable DataScaler for that chat.
DataScaler publishes a dynamic client registration endpoint, allowing Claude to create an OAuth client during setup. Free, Pro, and Max users can add a connector directly; Free accounts are currently limited to one custom connector. Team and Enterprise workspaces generally require an administrator to add the connector at the organization level before members authorize their own accounts. Menus and plan limits can change, so check Anthropic's custom connector guide for the current flow.
Claude Code
Claude Code supports remote HTTP MCP servers. Use claude mcp add:
claude mcp add --transport http datascaler https://app.datascaler.ai/mcp
Check the saved connection:
claude mcp list claude mcp get datascaler
Then enter:
/mcp
Claude Code opens the OAuth flow in a browser. After authorization, return to Claude Code and verify the connection. There is no need to copy an OAuth token into the terminal or store an API key in the project.
For a shared Claude MCP config, add .mcp.json at the project root:
{ "mcpServers": { "datascaler": { "type": "http", "url": "https://app.datascaler.ai/mcp" } } }
This file stores the server definition only. Each team member still authorizes access with their own DataScaler account. Other MCP clients may use different fields; see the DataScaler MCP documentation before reusing the configuration elsewhere.

Verify the connection
For the first test, use a brand the team already knows and specify the time range and required output:
Review public discussions about Bambu Lab from the past 30 days. Return the change in mention volume, the three most discussed topics, and the main negative feedback. Include sources for each finding.
Check four items:
- Claude called DataScaler rather than answering from general knowledge.
- The brand and time range are correct.
- The response includes sources that can be opened.
- The current DataScaler account has access to the brand.
After the first response, narrow the request:
- Which platform contributed most of the change?
- Which negative theme increased the most?
- Is the discussion about the product, price, delivery, support, or campaign execution?
- Which brands appear in the same purchase decision?
- Return three source posts that best support the finding.
Specific requests are easier to verify. Include at least the brand, time range, analysis dimension, and whether sources are required.
Three common query patterns
Brand changes
Compare the past seven days with the previous seven days. List the three largest changes for [brand] and identify the related platforms and sources.
This pattern can support a weekly review or campaign analysis. Percentage changes should be read with the underlying volume and sources; a percentage alone is not a business conclusion.
Competitor context
Find public discussions that mention [brand] and [competitor]. Organize them by purchase context, price, product capability, and customer concern, with source links.
Co-mentions do not always indicate direct competition. Sponsorship disclosures, tags, and incidental references should be separated. For a fuller screening process, use the direct competitor identification framework.
Customer questions and content planning
Find questions buyers repeatedly ask before purchasing [category]. Separate comparison, validation, and usage-risk questions, and include representative sources.
These questions can inform an FAQ, product page, or content plan. For the full workflow, see how teams use DataScaler MCP to find buyer-intent content ideas.
MCP, the DataScaler app, and Playground
Interface | Primary use |
|---|---|
DataScaler MCP | Querying, following up, comparing, and returning sources from Claude |
DataScaler app | Configuring brands and competitors, reviewing the full dashboard, filtering posts, and managing reports |
Playground | Previewing brand keywords, competitors, and available public discussions before creating a report |
The three interfaces serve different purposes. MCP supports analysis from Claude. The app remains the place for configuration and full data review. Playground provides an initial preview.
Data and product limits
- Results are limited to brands the current DataScaler account and organization can access.
- Data comes from public social, community, and review sources covered by DataScaler.
- Results depend on the selected period, platform coverage, and collection status.
- Sentiment and topic labels support filtering and analysis; they are not a survey of all consumers.
- The MCP service does not publish, comment, or reply on a brand's behalf.
- Before citing a finding externally, open the source and confirm the original context.
Frequently asked questions
Can Claude Code use MCP?
Yes. Run claude mcp add --transport http datascaler https://app.datascaler.ai/mcp, then use claude mcp list, claude mcp get datascaler, and /mcp to verify the configuration and complete OAuth.
Where should the Claude MCP config live?
For individual use, the claude mcp add command is sufficient. Use a project-level .mcp.json when a team needs to share the server definition. Include type: "http" and the server URL, but do not commit a personal token.
Should I use OAuth or an API key?
Claude supports OAuth, so OAuth is the default option. An API key is appropriate for clients or scripts that cannot complete OAuth. Store it in a secret manager or environment variable.
Why is Claude not calling DataScaler after it connects?
Check /mcp or Connectors first. Then confirm that the request includes a brand, time range, and query goal. DataScaler cannot return a brand that has not been configured, that the current account cannot access, or when the connector is disabled for the conversation.
Connect an account or preview a brand
Existing DataScaler users can start from the DataScaler MCP page. If no brand report exists yet, enter a brand or website in Playground to preview the keywords, competitors, and public discussions DataScaler can identify.
This guide was checked against DataScaler and Anthropic documentation in October 2026. Client menus, plan limits, and authentication flows may change; use the current DataScaler and Claude documentation for the latest configuration.
