Connect Fiserv to Claude: Automate Checkouts and Billing Cycles

If your team uses ChatGPT, check out our guide on /connect-fiserv-to-chatgpt-manage-payments-disputes-and-tokens/ or explore our broader architectural overview on /connect-fiserv-to-ai-agents-scale-payments-and-financial-adjustments/.

If you need to connect Fiserv to Claude to automate checkout flows, manage complex dispute cases, or scale OmniPay financial adjustments, you need a Model Context Protocol (MCP) server. This server acts as the critical translation layer between Claude's LLM function calls and Fiserv's vast, heavily regulated API surface. You can either build and maintain this infrastructure yourself, or use a managed integration platform like Truto to dynamically generate a secure, authenticated MCP server URL.

Giving an AI agent read and write access to a global payment processor like Fiserv is an engineering challenge. You have to handle payment card tokenization lifecycles, map massive transaction schemas to MCP tool definitions, and deal with Fiserv's strict multi-institution credential requirements. Every time Fiserv updates a schema or enforces a new security standard, you have to update your server code, redeploy, and test the integration.

This guide breaks down exactly how to use Truto to generate a secure, managed MCP server for Fiserv, connect it natively to Claude Desktop, and execute complex payment and billing workflows using natural language.

The Engineering Reality of the Fiserv API

A custom MCP server is a self-hosted integration layer. While the open MCP standard provides a predictable way for models to discover tools, the reality of implementing it against a massive financial API like Fiserv is painful. Fiserv is built to handle global transaction volume, multi-currency processing, and complex dispute lifecycles. Its API architecture reflects that complexity.

If you decide to build a custom Fiserv MCP server, here are the specific integration challenges you will face:

Fragmented Endpoints and Multi-Institution Credentials Fiserv is not a single API. Operations like creating a checkout session, submitting an OmniPay financial adjustment, and managing a chargeback dispute often span different host environments and require distinct credential payloads. For example, when querying the Disputes API (list_all_fiserv_disputes), you must pass a specific Institution number in the credentials if your API key covers multiple institutions. OmniPay Financial Adjustments require their own environment flags to ensure test keys route correctly to the CAT host. An LLM has no context on these routing rules. You must build an abstraction layer that handles these credential injections seamlessly.

Complex Batch Constraints on Financial Adjustments Fiserv enforces strict rules on how adjustments are applied to merchant accounts. The OmniPay Financial Adjustments API allows you to submit arrays of fees and adjustments, but limits this to 1,000 items per call. Furthermore, modifications (fiserv_modify_adjustments_bulk_update) can only be applied to adjustments that have not yet been picked up for processing by the daily cycle (their status must not be 'Released' or 'Cancelled'). Exposing these batch constraints directly to an LLM will result in failed validation loops. Your MCP tools must explicitly define these schema limits and state constraints so Claude knows exactly what is allowed.

Rate Limits and 429 Errors Financial APIs enforce aggressive rate limiting to protect their infrastructure. Truto does not retry, throttle, or apply backoff on rate limit errors. When the Fiserv upstream API returns an HTTP 429, Truto passes that error directly back to the caller. Truto normalizes the upstream rate limit information into standardized headers (ratelimit-limit, ratelimit-remaining, ratelimit-reset) per the IETF spec, but the caller is fully responsible for retry and backoff logic. Do not expect your MCP server to magically absorb rate limit errors - your agentic workflows must be engineered to handle 429s gracefully.

How to Generate the Fiserv MCP Server

Truto handles the complexity of the Fiserv API by dynamically generating MCP tools based on the integration's OpenAPI specifications and documentation records. Tools are not cached or hand-coded; they are generated on the fly when Claude requests them, ensuring your agent always has access to the most up-to-date schema definitions.

To connect Claude to Fiserv, you first need to generate a self-contained MCP server URL. This URL contains a cryptographic token that securely identifies the specific Fiserv account and the allowed toolset. You can do this via the Truto UI or programmatically via the API.

Method 1: Creating the Server via the Truto UI

If you are setting this up manually for an internal operations agent, the Truto dashboard is the fastest route:

1. Log into your Truto account and navigate to your connected Fiserv integrated account.
2. Click on the MCP Servers tab.
3. Click Create MCP Server.
4. Configure the server. You can limit access to specific methods (e.g., only read operations) or apply tag filters (e.g., only tools tagged with disputes).
5. Click Save, and copy the generated MCP server URL (it will look like https://api.truto.one/mcp/abc123def456...).

Method 2: Creating the Server via the API

If you are building an AI product and need to provision Fiserv tools for your users dynamically, you can use the Truto API to generate the server URL on the fly. Send a POST request to the /integrated-account/:id/mcp endpoint.

curl -X POST https://api.truto.one/integrated-account/<your_fiserv_account_id>/mcp \
  -H "Authorization: Bearer <your_truto_api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Fiserv Dispute Management AI",
    "config": {
      "methods": ["read", "write", "custom"],
      "tags": ["payments", "disputes"]
    },
    "expires_at": "2027-01-01T00:00:00Z"
  }'

The Truto API will validate the configuration and return a secure JSON-RPC endpoint:

{
  "id": "mcp_8a9b0c1d",
  "name": "Fiserv Dispute Management AI",
  "config": {
    "methods": ["read", "write", "custom"],
    "tags": ["payments", "disputes"]
  },
  "expires_at": "2027-01-01T00:00:00Z",
  "url": "https://api.truto.one/mcp/f8e9d0c1b2a3..."
}

Connecting the Fiserv MCP Server to Claude

Once you have your Truto MCP server URL, you must register it with Claude. Claude communicates with the server using standard JSON-RPC 2.0 messages over HTTP POST.

Method A: Connecting via the Claude UI

If you are using a managed Claude workspace or ChatGPT for your enterprise environment, you can connect the URL directly in the application settings:

1. Open your AI client settings (e.g., in Claude: Settings -> Integrations -> Add MCP Server).
2. Provide a recognizable name (e.g., "Fiserv Production").
3. Paste the URL you generated from Truto (https://api.truto.one/mcp/abc123def456...).
4. Save the configuration. Claude will immediately call the /tools/list endpoint and populate the agent's context window with the available Fiserv operations.

Method B: Connecting via Manual Configuration File

If you are running Claude Desktop locally or orchestrating custom AI agents, you can mount the MCP server using Server-Sent Events (SSE). Update your claude_desktop_config.json file to instruct Claude to route tool calls to Truto using the official @modelcontextprotocol/server-sse package.

{
  "mcpServers": {
    "fiserv": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse",
        "https://api.truto.one/mcp/f8e9d0c1b2a3..."
      ]
    }
  }
}

Restart Claude Desktop. The agent is now fully authenticated and ready to interact with your Fiserv environment.

Hero Tools for Fiserv

Truto dynamically generates a massive suite of tools covering the entire Fiserv REST API. To keep your agents highly focused, we recommend filtering your MCP server down to specific operational domains. Here are the highest-leverage tools for automating checkouts, managing disputes, and processing financial adjustments.

create_a_fiserv_payment

This is the core tool for initiating primary transactions in Fiserv, such as sales, pre-authorizations, or credit refunds. Claude must provide the requestType and the transactionAmount object to execute the call.

"Process a new sale transaction for $250.00 USD on the terminal. Ensure the request type is set correctly for an immediate capture."

get_single_fiserv_payment_by_id

When investigating order failures or confirming if a transaction settled successfully, Claude uses this tool to check the current state of an existing payment. It returns over 30 fields, including the transactionStatus, approvalCode, and paymentMethodDetails.

"Look up payment ID 'pay_987654321' and tell me if the transaction status is approved or declined. If declined, give me the exact transaction result code."

list_all_fiserv_disputes

Managing chargebacks manually is a massive drain on operational teams. This tool allows Claude to query open disputes by date range, merchant ID, or acquirer reference number. Note that if your API key spans multiple institutions, the Institution number must be configured in your underlying connection.

"Fetch all OPEN disputes logged in the last 7 days for Merchant ID 445566. Summarize the total disputed amounts and list their stage codes."

create_a_fiserv_dispute_note

During a chargeback investigation, teams need to log evidence and operational context. This tool attaches a text note to an existing dispute case. Once added, notes cannot be updated, so Claude must formulate the content precisely.

"Add a note to dispute ID 'dsp_112233' stating: 'Customer contacted support on Oct 12th and confirmed receipt of the digital goods. Logs attached in external system.'"

create_a_fiserv_post_adjustment

This is a critical tool for OmniPay environments. It allows the agent to submit an array of fees or financial adjustments against merchant accounts. It accepts up to 1,000 adjustments per call, which will be applied during the next merchant payment cycle.

"Create a post adjustment for merchant account 'merch_8899' applying a $15.00 USD fee rebate for their recent volume discount. Set the adjustment category to 'FEE_REBATE'."

fiserv_modify_adjustments_bulk_update

If a mistake is made before the daily processing pickup, Claude can modify existing OmniPay adjustments. This tool requires a JSON array of the modifications. Adjustments can only be modified if their status is neither 'Released' nor 'Cancelled'.

"Update the amount for adjustment slip number 'adj_5544' to $20.00 USD. Verify that the adjustment is still in a pending state before applying the modification."

To view the complete inventory of Fiserv tools, including payload schemas for token verifications, checkout sessions, and payment links, check out the Fiserv integration page.

Workflows in Action

Once connected, Claude can orchestrate complex, multi-step workflows across the Fiserv API. The flat input namespace of the MCP tools means Claude handles the parameter routing automatically, determining which arguments belong in the query string and which belong in the JSON body based on the schemas provided by Truto.

Scenario 1: Triaging and Updating Chargeback Disputes

Managing chargebacks requires speed. In this workflow, the operations manager asks the agent to review recent disputes and update a specific case with internal context.

"Pull our open Fiserv disputes from the last 48 hours. Find the one related to transaction 'txn_999888'. If it is in the 'PRE_ARBITRATION' stage, add a note to the case saying we are preparing the delivery confirmation documents."

1. Query Disputes: Claude calls list_all_fiserv_disputes with the status set to OPEN and the requested date bounds.
2. Identify the Case: The LLM parses the returned JSON array, matching the specific acquirerReferenceNumber or originalTransactionId to find the correct dispute ID.
3. Assess the Stage: Claude reads the stageCode field on the specific dispute.
4. Log the Note: Since the case matches the criteria, Claude invokes create_a_fiserv_dispute_note passing the dispute_id, a required tcn (transaction control number), and the provided text content.

sequenceDiagram
    participant User as Operations Manager
    participant Claude as Claude Desktop
    participant MCP as Truto MCP Server
    participant Upstream as "Upstream API (Fiserv)"

    User->>Claude: "Pull open disputes and note txn_999888 if pre-arb..."
    Claude->>MCP: Call list_all_fiserv_disputes (status: OPEN)
    MCP->>Upstream: GET /disputes?status=OPEN
    Upstream-->>MCP: [Dispute Array]
    MCP-->>Claude: JSON Tool Result
    
    Note over Claude: Parses array to find txn_999888<br>Checks stageCode == PRE_ARBITRATION

    Claude->>MCP: Call create_a_fiserv_dispute_note (dispute_id, text)
    MCP->>Upstream: POST /disputes/{id}/notes
    Upstream-->>MCP: 201 Created (Note ID)
    MCP-->>Claude: JSON Tool Result
    Claude-->>User: "Note added successfully to dispute dsp_123."

Scenario 2: Processing OmniPay Fee Adjustments

Finance teams frequently need to adjust fee schedules or apply manual credits for high-volume merchants. This workflow automates bulk OmniPay post adjustments.

"We need to apply a $50.00 USD manual hardware credit to merchant 'M-4455'. Check if they have any pending unreleased adjustments first. If not, create the new post adjustment."

1. Check Staged Adjustments: Claude calls list_all_fiserv_get_adjustments_stageds filtering by the merchantId to retrieve any pending adjustments that haven't hit the daily cycle.
2. Analyze the Results: The LLM determines if any existing adjustments match the criteria or conflict with the request.
3. Create the Adjustment: Assuming the queue is clear, Claude calls create_a_fiserv_post_adjustment, formatting the JSON array correctly for the OmniPay environment, injecting the merchant ID, the amount, currency, and the adjustmentType.
4. Report Back: Claude confirms the adjustmentSlipNumber returned by Fiserv to the finance user.

Security and Access Control

Exposing financial APIs to AI agents requires strict access controls. Truto provides multiple layers of security at the MCP token level to ensure your Fiserv data remains protected:

• Method Filtering: Use the config.methods array when creating the MCP server to restrict the LLM to specific operations. For example, passing ["read"] ensures Claude can only call list or get endpoints, completely preventing the agent from initiating payments or modifying disputes.
• Tag Filtering: Use the config.tags array to scope the server to a specific domain. Passing ["disputes"] will expose only the dispute-related tools, hiding checkout, OmniPay, and tokenization endpoints from the agent entirely.
• Require API Token Auth: By default, the MCP server URL contains a cryptographic hash that serves as authentication. For high-security financial workflows, you can set require_api_token_auth: true. This forces the client to pass a valid Truto API token in the Authorization header, meaning possession of the URL alone is useless without valid backend credentials.
• Expiration (TTL): You can set an expires_at timestamp when generating the server. Once this time is reached, durable alarms automatically clean up the underlying database records and KV entries, instantly invalidating the MCP server and terminating the agent's access to Fiserv.

Moving Forward with Agentic Payments

Connecting Fiserv to Claude transforms complex, multi-system financial operations into conversational workflows. Instead of forcing support and finance teams to manually click through dashboards to investigate a chargeback or process an OmniPay adjustment, you can empower AI agents to query the state, evaluate the rules, and execute the exact API calls required.

Building this orchestration layer from scratch means fighting with multi-institution credentials, batch array limits, and strict IETF rate limit headers. By leveraging a managed MCP server, you offload the infrastructure, security, and schema management entirely. You simply point Claude at the endpoint, and the tools work.