Guides

Connect Contend via MCP

Expose your competitive intelligence to ChatGPT, Claude Desktop, and other LLM clients via the Contend MCP server

Contend exposes a Model Context Protocol server so LLM clients — ChatGPT, Claude Desktop, and any other MCP-capable agent — can read your competitive intelligence directly. Your competitors, their events feed, pricing, and feature classifications are all available to a connected LLM as structured tools rather than a screenshot-and-paste workflow.

What it gives you

When an MCP client is connected to Contend, the LLM can:

  • List and manage your competitors. Query the monitored list, look up a specific competitor, or add new competitors by URL (up to 50 at a time, with auto-enrichment and tagging).
  • Read the events feed. List recent company change events across your monitored competitors — filter by date range, importance level (low / medium / high), event type, and competitor id — and fetch the full body of any single event (blog post, changelog entry, and so on) on demand.
  • Inspect competitor pricing. Retrieve the active pricing model for any monitored competitor — tiers, plans, and price points — ready for comparison.
  • Work with your feature taxonomy. Read the feature vocabulary built from your website, and the per-competitor feature classifications that back the Features comparison report.

All of it is scoped to your account. The LLM sees what you see in the app, nothing else.

How authentication works

You sign in through your browser. No API keys to generate, paste, or store.

When you add Contend to your MCP client, it opens the Contend sign-in page, you pick an account and click Approve, and that's it — you stay signed in for about 30 days before being asked to re-authorize. Your client handles everything else; the only thing you ever paste is the server URL.

Server URL

https://api.contend.so/mcp/contend

That is the only URL you need. Paste it into your client's "Add connector" or "Add MCP server" dialog; the client handles the rest.

Connecting ChatGPT

Custom MCP connectors in ChatGPT require Developer Mode, which is available on Business, Enterprise, and Edu plans.

  1. Enable Developer Mode: open Settings → Apps & Connectors → Advanced settings, then toggle Developer mode on.

  2. Go to Settings → Connectors → Create, paste https://api.contend.so/mcp/contend as the connector URL, give it a name and description, and submit.

  3. ChatGPT will open a Contend sign-in window — sign in, choose your account, and click Approve.

  4. Back in ChatGPT, the Contend tools appear in the tool picker. Try:

    What changed in our top five competitors this week? Focus on high-importance signals.

    ChatGPT will call list_events under the hood and return a grounded summary.

OpenAI is renaming "Connectors" to "Apps" (announced December 17, 2025), so menu labels may have shifted by the time you read this — the flow is the same.

Connecting Claude Desktop

  1. In Claude Desktop (or at claude.ai), open Settings → Connectors, scroll to the bottom, and click Add custom connector.

  2. Paste https://api.contend.so/mcp/contend as the server URL and click Add.

  3. Claude opens a Contend sign-in window — sign in, choose the account you want to authorise, and click Approve.

  4. Back in Claude, the Contend tools appear in the tool picker and can be called from any conversation. Try:

    Summarise pricing changes across my monitored competitors in the last 30 days.

    Claude will call get_pricing and list_events under the hood.

Custom connectors are currently in beta and available across Claude's Free (one connector), Pro, Max, Team, and Enterprise plans. No config file editing — the Connectors UI handles registration and OAuth end-to-end.

Tools available

Competitors are addressed by id everywhere. list_competitors returns each competitor's id, which is then the handle for get_competitor, get_pricing, get_competitor_features, and the competitor_ids filter on list_events. Events are addressed by event_id.

Grouped by toolset:

contend_competitors

  • list_competitors — List your monitored competitors. Supports tag filtering and pagination.
  • get_competitor — Retrieve a single competitor by id.
  • add_competitors — Batch-add up to 50 competitors by URL. Auto-enriches the company profile and can apply tags at creation.

contend_events

  • list_events — List recent company change events across your monitored competitors. Filter by date range (since / until, ISO-8601), importance threshold, event_types, and competitor_ids. Returns lean event metadata by default; set hydrate=true to include the full object body (blog post, changelog entry, etc.) for every result in one call. Paginated via limit and offset.
  • get_event — Fetch a single event by event_id. Hydrated by default (returns the full object body); set hydrate=false if you only need the metadata.

contend_pricing

  • get_pricing — Active pricing model for a monitored competitor by competitor_id, including tiers and price points.

contend_features

  • get_feature_taxonomy — Your account's feature vocabulary (categories and non-deprecated feature entries).
  • get_competitor_features — Feature classifications for a specific competitor by competitor_id.

All tools return JSON. Errors include error, tool, message, and type fields so the LLM can recover or explain the failure clearly.

What makes this useful

The canonical MCP use case for Contend is replacing the "open the app, scan the events feed, summarise for the team" workflow with a natural-language one. A connected LLM can answer:

  • "Summarise competitor moves from the last 7 days at high importance only."
  • "Which of our monitored competitors changed pricing in Q1?"
  • "Compare our top three competitors' feature coverage for workflow_automation."
  • "Add these five URLs as competitors and tag them emerging."

Because the LLM is operating on live data through structured tools, the answers are grounded in your actual Contend state — not a scrape, not a screenshot, not last month's export.

Revoking access

Removing the connector from your client ends the session. To revoke immediately, contact the team via the contact page. Self-serve connection management is on the roadmap.

  • Introduction — how Contend captures the signals the MCP server exposes.
  • General FAQ — plans, data sources, and delivery channels.