MCP documentation

All Ears MCP server

All Ears provides a hosted Model Context Protocol (MCP) server that gives AI assistants direct access to your All Ears Monitors, saved alerts that continuously collect and enrich mentions of the keywords you care about across podcasts, radio, YouTube, TikTok, TV, and news, in 11 languages.

Overview

Once connected, you can list your monitors, pull enriched mentions and daily metrics, break down sentiment, and run trend/spike/growth/top-source analytics, all through natural conversation, no dashboard navigation required.

A typical workflow looks like:

1. Discover: list your monitors, then choose which one(s) you want to continue working with

2. Read: pull recent mentions, daily metrics, or sentiment for that monitor

3. Analyze: trend, compare, or break down coverage across channels

4. Iterate: refine with follow-up questions in the same conversation

A monitor mirrors what's already available in your All Ears account: it's the same saved alert you'd see in the All Ears UI, continuously collecting and enriching mentions of the keywords you configure. Monitors you create through MCP appear in your All Ears account just like ones created in the UI / platform, and vice versa: if your account has access to more monitors than the ones you've already used with your assistant, you can create new ones through either MCP or the UI and they'll show up in both places.

Available tools

Overview

Creating a monitor is a background job. create_monitor returns immediately, but historical mentions are backfilled asynchronously, and sentiment is computed by a separate process after mentions land. For a broad topic this can take minutes to hours, with no progress indicator to poll, so just re-check later and sentiment will fill in from empty.

• Every mention includes sentiment and can be filtered by sentiment, reach, and free text.

• Episodes vs. mentions. A monitor distinguishes distinct episodes (a podcast episode, a broadcast, a post) from individual mention passages within them: one episode can carry several mentions. Both counts are available.

• Up to 20 monitors can be compared in a single analytics call (e.g. your brand vs. five competitors), with one row per monitor in the response.

• New monitors can be created up to your account's monitor limit. create_monitor through MCP counts against the same limit as monitors created in the All Ears UI. If you're at your limit, the call returns an error instead of creating the monitor. Contact your account manager or support@allears.ai to raise your limit.

MCP Server URL

Transport: Streamable HTTP. Authentication: OAuth, handled automatically by your client. You log in with your normal All Ears credentials once in the browser.Contact us

Permissions & access

MCP is included in your paid All Ears plan and gives you access to the Monitors included in your plan.

The first time your assistant calls an All Ears tool, a browser window opens for you to log in and approve access. After that, your client manages authentication automatically.

Any user on your organization's All Ears account can connect once your account manager has enabled the Monitors product. Users authenticate with their own credentials, so existing account permissions apply.

Connecting to the MCP Server

Claude (Desktop & webb)

Go to Settings → Connectors → Add custom connector, name it "All Ears", and paste the server URL https://mcp.allears.ai. Claude opens a browser window for you to log in the first time you use a tool.

Claude Code

Shell: claude mcp add --transport http allears https://mcp.allears.ai

Cursor

Open Settings → MCP → Add new MCP server, choose type Streamable HTTP, and use the URL https://mcp.allears.ai. Or add this to your mcp.json:

{

  "mcpServers": {

    "allears": {

      "url": "https://mcp.allears.ai"

    }

  }

}

ChatGPT

In a Developer mode / connectors-enabled workspace, add a new connector pointing at https://mcp.allears.ai and complete the OAuth login when prompted.

Other clients

Any MCP client that supports remote servers over Streamable HTTP with OAuth will work. Point it at https://mcp.allears.ai. Most clients accept a config block like the Cursor example above.

Example Queries

Once connected, try asking your AI assistant:

Discover

• "Show me all my monitors."

• "What keywords is my 'Climate Change' monitor tracking?"

Read mentions & metrics

• "What were the latest mentions for my 'Climate Change' monitor?"

• "Give me daily reach and mention counts for that monitor last month."

Sentiment

• "How did sentiment split for that monitor last week?"

• "What's the net sentiment trend for our brand monitor this quarter?"

Coverage & channels

• "Which channel types is that monitor picking up the most coverage on?"

• "Who are the top sources covering this topic?"

• "Are there any spikes in coverage I should know about?"

Compare

• "Compare episode count for us vs. our three competitor monitors last month."

Create & manage

• "Create a new monitor tracking 'electric vehicles' and 'EV' in English, across podcast and news."

• "Add 'battery range' as a keyword to my EV monitor."

Security Considerations

When connected, the AI assistant can read your monitor configuration and data, and, if granted monitor:write, create or edit monitors on your behalf. Your All Ears data is also sent to whichever AI provider you're using, so review that provider's data handling policy before connecting.

Access controls

• MCP access follows the same account-level Monitors entitlement as the REST API

• Users can only access monitors already visible to their All Ears account

• Existing account-level permissions and roles remain in effect

Other considerations

• If using a shared AI workspace, be aware that conversation history may be visible to others on that workspace

• Sentiment on a newly created monitor is incomplete until the async backfill finishes. The assistant should tell you this rather than presenting partial results as final

Troubleshooting

• "Limit reached" when creating a monitor: your account is at its monitor limit. Archive a monitor you no longer need, or contact your account manager or support@allears.ai to raise your limit.

• Sentiment fields are null or incomplete: the monitor was created recently and its backfill hasn't finished. Re-query after a few minutes to a few hours depending on topic volume.

• Client doesn't prompt for login: some clients cache a stale token. Remove and re-add the connector, or clear your client's MCP auth cache and reconnect.

• Can't find a monitor by name: call list_monitors first to resolve the correct UUID, since monitor titles aren't always unique identifiers across tools.

Questions or access requests? Contact support@allears.ai.