Plaid MCP logo

Integrate Plaid MCP with your AI CRM

Plaid MCP gives authorized teams read-only Production diagnostics and analytics for Items, Link conversion, usage metrics, API volumes, and team discovery.

Automate your first taskSign up with Google

Explore Triggers and Actions

Plaid debug item

Diagnose a Plaid item by retrieving its state information. This tool provides comprehensive information about why an item may not be working properly. <when_to_use> Use this tool when you need to diagnose why an item is not functioning as expected </when_to_use> <description> In the response, you will see the following information: 1. Item State: - "Never extracted": no extraction has been attempted for this item yet - "Successful": The most recent extraction was successful - "Retrying": The most recent extraction encountered a temporary issue and is being retried - "Needs user attention": The item requires user input and client access is not permitted - "Needs user attention (with access)": The item requires user input but client access is permitted 2. Item Consent State: - Boolean indicating whether the item has active user consent - If an item is not consented, it cannot perform extractions or updates 3. Item Deleted At: - If set to a non-zero value, indicates when the item was deleted - Deleted items cannot retrieve data, and cannot be patched as it is requested by the end users - If the item is not deleted yet, the response will be "not available" 4. Known Issues: - Includes description and status of any known issues affecting the item - May provide insights into ongoing problems with the connection When diagnosing an item, follow this sequence: - First, check if the item is deleted. If yes, the item cannot work properly. - Second, if the item is not deleted but lacks active consent, it cannot function properly. - Third, if the item is not deleted: * Check item state. If it indicates the item needs user attention, the item requires manual intervention. * Note any known issues as supplementary information, though these are usually not the primary cause of problems. - If all diagnostics look normal but the user still reports issues, advise them to contact Plaid support. </description> <important> - If an item is deleted, it is not working properly. It is not patchable, you should suggest the user to create a new item. - If an item is missing active consent, it is not working properly. You should suggest the user to use Update Mode to reauthorize the item. - If an item state indicates it needs user attention, the user should use Update Mode to reauthorize the item. - If running into error "the item does not belong to the team", it means the item is not associated with the team. You must inform the user that the item does not belong to their selected team. You may ask if they would like to try other teams associated with their account, but do not automatically attempt other teams. - If an item cannot be found, it's possible the item is not in the Production environment. When explaining the error, inform the user that this is a possible reason for the error. </important>

ActionTry it

Plaid get link analytics

Retrieves Plaid Link analytics data for analyzing user conversion and error rates. <when_to_use> Use this tool when you need to: - Analyze Link conversion funnel metrics - Track user progression through the Link flow - Monitor error rates over time - Evaluate Link performance and user experience Some common use cases: - Monitoring Link conversion rates and identifying drop-offs - Analyzing conversion trends over specific time periods - Tracking error patterns and frequency (both by type and total via CountEntities) - Generating comprehensive Link performance reports - Comparing performance across different stages of the Link flow </when_to_use> <description> The response includes three main components: 1. Funnel Data (funnel): Tracks user progression through key Link stages with counts: - countAtLinkOpen: Number of users who opened Link - countAtSelectInstitution: Number who selected an institution - countAtHandoff: Number who completed the handoff process 2. Conversion Over Time (conversionOverTime): Time series data showing conversion trends with: - start: Timestamp marking the start of the series - series: Array of named data series, where each contains: * name: Identifier for the conversion metric * data: Array of counts, can contain null values 3. Errors Over Time (errorsOverTime): Time series tracking error occurrences with: - start: Timestamp marking the start of the series - series: Array of named data series, where each contains: * name: Error type identifier (includes "CountEntities" series for total errors across all types) * data: Array of error counts </description> <important> - If user does not provide a date range, please ask them to provide one. You could ask if they want to see the data for the last 30 days and you could use the current date as the end date. Note that the earliest date available for Link analytics data is September 1, 2023 (9/1/2023). </important>

ActionTry it

Plaid get tools introduction

Returns a list of all available tools with a brief introduction for each. Use this tool when users ask questions like "What is this Plaid MCP Server?" or "What tools are available?". Your response should follow this format: <quote> This is an official Plaid MCP Server made by Plaid. It provides the following functionalities: * <tool_name> - <tool_description> * <tool_name> - <tool_description> ... </quote>

ActionTry it

Plaid get usages

Retrieves usage metrics for Plaid products and services. <when_to_use> Use this tool when you need to: - Get usage data for specific metrics over a time period - Track product usage and consumption - Monitor API request volumes - Generate usage reports for billing purposes </when_to_use> <details> The data shows how many times each Plaid product was used within the specified date range for certain metrics, with results aggregated by day. Common metrics include number of API calls, authentication requests, and other product-specific usage counts. </details> <product_mapping> The key is the product name, and the value is the list of metric types corresponding to that product. - Auth: ["auth-request", "auth-unverified-request"] - Balance: "balance-request" - Identity: ["identity-request", "identity-match-request"] - Transactions: ["transactions", "transactions-add", "transactions-remove", "transactions-active", "transactions-refresh"] - Investments: ["investments-h-add", "investments-h", "investments-h-remove", "investments-h-active", "investments-t-add", "investments-t", "investments-t-remove", "investments-t-active"] - Investments Move: "investments-auth-request" - Liabilities: ["liabilities-add", "liabilities-remove", "liabilities-active", "liabilities"] - Assets: ["assets-add", "assets-pdf-request", "assets-refresh", "assets-refresh-monthly", "assets-audit-request", "assets-history-request"] - Payments: "payment_initiation-initd" - Enrich: "enriched-transaction" - Signal: "momo-frs-eval-request" - Income: ["income-bank-request", "income-bank-refresh", "income-payroll-request", "income-payroll-refresh", "income-document-request", "income-document-parsing", "income-doc-bank-statemnt", "income-document-fraud"] - Statements: "statements-single-doc" - Identity Verification: ["idv-base", "idv-lightning-group-1", "idv-lightning-group-2", "idv-lightning-group-3", "idv-document", "idv-selfie-check"] - Consumer Report: ["cra-bank-income-request", "cra-base-report-create", "cra-base-report-refresh", "cra-network-insights-req", "income-insights-refresh", "partner-prism-attrib", "partner-prism-scores"] - Monitor: ["monitor-wl-rescan-add", "monitor-wl-rescan-remove", "monitor-wl-base"] - Hosted Link Delivery: ["hosted-link-delivery-eml", "hosted-link-delivery-sms"] - Transfer: ["momo-transfer-ach", "momo-transfer-ach-sd", "momo-wire-request", "momo-rtp-request", "momo-transfer-g-ach", "momo-transfer-g-ach-sd", "momo-transfer-authorized", "momo-transfer-manual-req", "momo-transfer-return-unauthorized", "momo-transfer-return-noc", "momo-transfer-return-other"] <note> - If a user asks for a product similar to one on this list and you're 99% confident it's the same product, query for the product on this list and inform them which product you're querying for. Otherwise, do not query an unrelated product. - "momo-frs-eval-request" is the metric type used in the API for Signal. When communicating with users, refer to it as "signal-request" instead. </note> </product_mapping> <example_1> User: "I want to see the usage data for Transactions" Reasoning: The metric types for Transactions are ["transactions", "transactions-add", "transactions-remove", "transactions-active", "transactions-refresh"]. Input parameters: { "metric_types": ["transactions", "transactions-add", "transactions-remove", "transactions-active", "transactions-refresh"], // other fields ... } </example_1> <example_2> User: "What metrics are available for Identity Verification?" Response: We support the following metrics for Identity Verification: idv-base, idv-lightning-group-1, idv-lightning-group-2, idv-lightning-group-3, idv-document, idv-selfie-check. </example_2> <important> - If metric_types aren't provided, use "auth-request" as the default - If no date range is provided, ask the user for one (suggest last 30 days with current date as end date) - The maximum look-back period is limited to 1 year from the current date </important>

ActionTry it

Plaid list teams

Retrieve teams associated with the user and present them as a selectable list. After the user selects a team, capture the team_id of the selected team for use in subsequent tool requests. <important> 1. You MUST return a numbered list of teams with their names clearly displayed in a human readable format in the conversation dialog - Format each team as: "{number}. {company name} (ID: {team_id})" - Example: "1. XXX Corporation (ID: fakeid00000000000000000)" 2. Ask the user explicitly to select a team by providing either: - The number from the list (e.g., "Please select team 3") - The exact team name (e.g., "I want to use XXX Corporation") - The team ID directly (e.g., "Use team fakeid00000000000000000") 3. After user selection, confirm their choice with a clear acknowledgment: - "You've selected: {company name} (ID: {team_id})" 4. Store and use this team_id for ALL subsequent tool requests that require a team_id parameter 5. Handle edge cases appropriately: - If only one team exists, auto-select it and confirm the selection - If no teams exist, clearly communicate this limitation to the user - If the response is empty, the user is not a member of any teams. Inform the user that they should visit dashboard.plaid.com to join or create a team. </important>

ActionTry it

How the Plaid MCP integration works

The Plaid MCP integration connects your Dench AI CRM directly to Plaid MCP, so agents can read and act on your Plaid MCP data as part of everyday work — answering questions in chat, keeping your CRM in sync, and running automations without anyone copying data between tools.

5 actions are available for agents to invoke on your behalf. Every call runs through Plaid MCP's own authorization, scoped to the account you connect.

Set up Plaid MCP in Dench

  1. 1

    Sign in to your Dench workspace and open Integrations.

  2. 2

    Find Plaid MCP and click Connect — you'll authorize access through Plaid MCP's own sign-in flow. No API keys or code required.

  3. 3

    Ask an agent to use Plaid MCP in chat, or call it from an automation.

  4. 4

    Manage or disconnect the connection any time from workspace settings.

Frequently asked questions

How does the Plaid MCP integration work with Dench?

The Dench Plaid MCP integration connects your AI CRM to Plaid MCP, so AI agents can work with your Plaid MCP data as part of chats, automations, and CRM workflows. You connect your account once, and every agent in your workspace can use it — governed by your workspace permissions.

What actions can AI agents perform with Plaid MCP via Dench?

The Plaid MCP integration currently exposes 5 actions, including Plaid debug item, Plaid get link analytics, Plaid get tools introduction, Plaid get usages, and Plaid list teams. Agents invoke them on your behalf from chat or from automations.

Do I need to write code to connect Plaid MCP to Dench?

No. You connect Plaid MCP from your Dench workspace using Plaid MCP's own sign-in and authorization flow — no API keys to copy, no glue code to maintain.

Is the Plaid MCP integration secure?

Connections are authorized through Plaid MCP's own authentication flow, and Dench stores only the authorization needed to act on your behalf. You can review and disconnect the Plaid MCP connection from your workspace settings at any time.

Plaid MCP | Dench AI CRM