Skip to main content

Talon.One MCP server Closed beta

The Talon.One MCP server connects your Talon.One environment to AI agents.

It gives agents secure, read-only access to your campaigns, customers, coupons, and loyalty programs. Agents provide answers about your promotions without you having to navigate the Campaign Manager or write code.

The Talon.One MCP server enables campaign managers, support agents, and anyone using AI agents to do the following, for example:

  • Answer questions in plain language about campaigns and loyalty programs.
  • Identify performance trends and pain points without building reports or navigating the Campaign Manager.
  • Investigate customer sessions and eligibility issues in seconds.

Supported tasks​

The Talon.One MCP server gives AI agents tools to read and understand your Talon.One data.

When connected, agents can, for example:

  • Explain the logic behind specific campaign rules and loyalty transactions.
  • Confirm if a campaign is live and check its current budget limits.
  • Analyze a customer's point balance, transaction history, and tier status to explain specific loyalty-related changes.
  • Investigate integration issues, such as API requests that failed or never reached Talon.One.
  • Provide accurate answers pulled directly from your live environment about current attributes, rules, campaigns, and limits.
Important

The Talon.One MCP server authenticates you through your Talon.One account, so agents access only the data allowed by your account permissions. Third-party AI services may process your data. Review your company AI policies before you connect tools that access customer data.

Prompt examples​

For marketing teams​

How is our summer sale campaign performing?

The AI agent can use the MCP server to:

  • Get redemption rates, budget usage, and loyalty points from live data.
  • Summarize referral conversions and audience engagement.

What targeting rules are configured on our VIP campaign?

The AI agent can use the MCP server to:

  • Explain the specific rules and audience configured for the campaign in plain language.
  • Provide accurate answers pulled directly from your live environment.

For support teams​

Why didn't this promotion apply for this customer?

The AI agent can use the MCP server to:

  • Evaluate the rule logic to find where the customer failed to meet requirements.
  • Identify if specific coupon limits or profile attributes blocked the promotion.

What's the loyalty status for customer X?

The AI agent can use the MCP server to:

  • Check the customer's point balance, transaction history, and tier status.
  • Provide an accurate answer pulled directly from your live environment.

For integration engineers​

Why is our integration returning errors?

The AI agent can use the MCP server to:

  • Check the Application health and review recently failed API requests in the access logs.
  • Identify patterns, such as authentication failures or incorrectly formatted requests.

For account administrators​

Who has access to this Application?

The AI agent can use the MCP server to:

  • List the users and roles, and explain each role's permissions in plain language.
  • Flag deactivated users that still hold active roles.

Supported MCP clients​

The Talon.One MCP server supports the following MCP clients:

  • Claude Desktop
  • Claude Code
  • Cursor
  • Microsoft Copilot
  • Copilot Studio
  • Gemini CLI
  • ChatGPT CLI
  • Codex CLI
  • Most other stdio-compatible clients

Prerequisites​

  • You have Node.js installed on your computer.
  • You have a Talon.One account. You sign in with your Talon.One credentials or Okta SSO.

Connect the MCP server​

Each Talon.One environment provides its own MCP server URL. The server uses OAuth 2.0, so you authorize access in your browser instead of managing an API key. To connect an agent, determine your MCP server URL, add the configuration to your client's config file, and authorize access when your browser opens.

Important

OAuth 2.0 replaces API key authentication for the Talon.One MCP server. API key authentication still works, but use OAuth 2.0 for new clients, and migrate existing clients. To migrate a client, replace its API key configuration with the OAuth 2.0 configuration in your AI client config file.

1

Determine your MCP server URL​

Your MCP server URL is based on your specific deployment address. To compose it, add /v1/mcp/entrypoint to the end of your deployment URL. Example: https://mycompany.europe-west1.talon.one/v1/mcp/entrypoint.

2

Configure your MCP client​

The Talon.One MCP server supports most stdio-compatible MCP clients. The following steps cover the most commonly used ones. For any other client, see your MCP client's documentation.

To connect Claude Desktop to the Talon.One MCP server:

  1. Open Settings > Developer.

  2. Click Edit Config to open claude_desktop_config.json.

  3. Add the following configuration to the file, replacing the URL with your MCP server URL:

    {
      "mcpServers": {
        "talon-mcp": {
          "command": "npx",
          "args": [
            "mcp-remote",
            "https://YOUR_DEPLOYMENT_URL/v1/mcp/entrypoint"
          ]
        }
      }
    }

  4. Save the file and restart Claude Desktop.

3

Authorize access in your browser​

After you save the configuration, the Talon.One sign-in page opens in your default browser. Sign in with your Talon.One credentials or Okta SSO to authorize access. Your MCP client loads the Talon.One tools after authorization completes.

Add skills​

Skills are instruction files that teach an AI agent how to use the Talon.One MCP server for a specific role. Each skill provides the agent with a tailored set of tools and step-by-step playbooks. This helps the agent choose the right tools and focus on the tasks that matter for that role.

To use a skill, download the file and add it to your MCP client's skills directory. For more information, see your MCP client's documentation.

The following skills are available:

  • Marketing manager: For reviewing, auditing, and analyzing campaigns, coupons, referrals, and giveaways at the campaign level.
  • Customer support agent: For per-customer investigations that start from a specific customer and ask why a process happens or fails.
  • Loyalty program manager: For working with loyalty data, such as point balances, tiers, ledgers, transactions, cards, and achievements.
  • Integration engineer: For debugging and auditing Talon.One at the API and configuration level, such as access logs, sessions, and webhooks.
  • Account administrator: For platform-level management across Applications, such as user access, roles, webhooks, and Application configuration.
On this page