Plaid MCP
Plaid MCP gives authorized teams read-only Production diagnostics and analytics for Items, Link conversion, usage metrics, API volumes, and team discovery.
Verdict
Common use cases
- Reconcile monthly expenses against bank statements
- Verify vendor payment status in real time
- Pull transaction history for audit prep
- Check account balances before approving transfers
- Identify duplicate charges across linked accounts
Integration
- Vendor
- Plaid MCP
- Category
- other
- Auth
- OAUTH2
- Composio slug
plaid_mcp
Tools
Setup
Setup guide
- 11. Create a Plaid developer account at dashboard.plaid.com and generate API keys for your environment (sandbox or production). 2. In Switchy, open Settings > Integrations and click Connect next to Plaid MCP. 3. Paste your Plaid client ID and secret when prompted, then authorize Switchy to access your Plaid data. 4. Plaid will redirect you through an OAuth consent screen—grant read access to transactions, balances, and identity scopes. 5. Back in Switchy, test the connection by creating a Project and typing '@Plaid list recent transactions' to confirm the AI can query your linked accounts. 6. To invoke Plaid tools in any Project, @mention Plaid followed by your request—the AI will route it through the MCP and return structured financial data.
What teammates see: by default, memories from Plaid MCP are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Reconcile Last Month
@Plaid retrieve all transactions from the last 30 days across linked accounts and highlight any charges over $500 or duplicate entries.
Verify Vendor Payment
@Plaid search for a payment to Acme Corp in the last 10 days and confirm the amount and settlement date.
Balance Check Before Transfer
@Plaid show current balances for all linked checking and savings accounts, then calculate total available funds.
Audit Trail Export
@Plaid export all transactions from January 1 to March 31 in a format I can import into Excel for audit.
Flag Unusual Activity
@Plaid analyze the last 60 days of transactions and flag any that are 3x higher than the account's average spend.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Plaid fetch my checking account balance and recent transactions from the last 7 days
Retrieved balance and transaction data for your Chase checking account (ending in 4892). Current balance: $3,247.18. Found 12 transactions in the past 7 days: 3 direct deposits totaling $2,100, 6 debit card purchases ranging from $8.50 to $127.43, 2 ACH transfers out totaling $450, and 1 ATM withdrawal of $60. All transactions cleared; no pending items.
This example shows read-only account access. Requires OAuth2 consent with 'transactions' scope. Plaid aggregates data from your linked bank; freshness depends on the institution's API update frequency (typically daily, some real-time). No write capability—Plaid cannot initiate transfers or payments.
@Plaid compare my spending on groceries vs dining out this month across all linked accounts
Analyzed transactions across 2 linked accounts (Chase checking, Amex credit). Groceries: 18 transactions totaling $612.34 (average $34.02 per trip). Dining out: 9 transactions totaling $387.50 (average $43.06 per meal). Grocery spending is 58% higher this month. Notable: 4 grocery trips over $50 suggest bulk shopping; dining transactions cluster on weekends.
This synthesis example pairs Plaid's transaction categorization with AI reasoning. Category accuracy depends on merchant data quality—some transactions may be miscategorized and require manual review. Requires 'transactions' scope across multiple accounts. Plaid does not store your prompts or analysis results.
@Plaid verify my income sources and employment history for the past 12 months
Income verification complete. Identified 2 income sources: Primary employer (Acme Corp) with 12 monthly deposits averaging $4,583.33, consistent timing on the 1st of each month. Secondary income (freelance payments via Stripe) with 7 deposits totaling $3,200 over the period. Employment tenure confirmed through continuous payroll deposits since January 2023. No gaps detected.
This example demonstrates Plaid's income and employment verification capabilities, often used for loan applications or rental screenings. Requires 'income' scope in addition to 'transactions'. Data accuracy depends on consistent deposit patterns—irregular income sources may need manual documentation. Plaid does not share this data with third parties without your explicit consent per request.
Use-case deep-dives
When Plaid MCP fits early-stage fintech user onboarding
A 6-person fintech startup runs user onboarding in Slack threads—support flags suspicious sign-ups, ops checks bank account status, product reviews friction points. Plaid MCP lets the team query account verification status and transaction history without leaving the workspace. OAuth2 means each team member authenticates once; the MCP pulls live data on demand. This works when your onboarding volume is under 200 accounts per week and you need spot-checks, not batch processing. If you're verifying thousands of accounts daily or need automated fraud scoring, you want Plaid's API in your backend, not an MCP in a chat workspace. For small teams doing manual review, this MCP turns a five-tab context switch into a single prompt.
Plaid MCP for support teams answering balance questions
A 4-person support team at a budgeting app fields 30-50 tickets daily asking 'why isn't my balance updating?' Plaid MCP gives support reps a way to pull the user's latest account sync status and transaction feed without admin panel access or engineering help. OAuth2 keeps credentials scoped per rep. This scenario wins when your support load is light enough that reps can handle queries conversationally—under 100 tickets per day, mostly one-off lookups. If you're running a high-volume support queue with SLA targets, you need Plaid data piped into Zendesk or Intercom, not fetched ad-hoc in a chat workspace. For small support teams, this MCP cuts median resolution time by skipping the 'ask engineering' step.
When Plaid MCP speeds up investor data requests
A 3-person finance team at a Series A startup gets weekly investor requests for connected account metrics—total linked accounts, transaction volume, institution breakdown. Plaid MCP lets the CFO or ops lead pull these snapshots in Switchy without writing SQL or waiting on engineering. OAuth2 means the finance team authenticates directly; no shared credentials. This works when you're fielding 5-10 ad-hoc data requests per month and the metrics are simple aggregates. If you're preparing board decks with complex cohort analysis or need automated monthly reports, you want Plaid data in your data warehouse, not queried through an MCP. For early-stage teams doing investor relations in real time, this MCP turns a two-day engineering ticket into a five-minute prompt.
Frequently asked
What does the Plaid MCP do in Switchy?
The Plaid MCP connects your team's financial accounts to Switchy's AI workspace through Plaid's banking API. Your AI agents can read transaction history, account balances, and institution details without manually exporting CSVs or logging into multiple banking portals. This works for checking accounts, credit cards, and investment accounts that Plaid supports.
Do I need special permissions to connect Plaid?
You'll go through Plaid's OAuth2 flow, which means you authenticate directly with your bank using your normal login credentials. Plaid never sees your password. The MCP requests read-only access to transactions and balances by default. If your bank requires multi-factor authentication, you'll complete that during setup. No admin access to Switchy itself is required.
Can the Plaid MCP initiate payments or transfers?
No. The integration is read-only for safety. Your AI agents can analyse spending patterns, reconcile transactions, or pull balance snapshots, but they cannot move money or modify account settings. If you need payment initiation, you'd handle that through your bank's own interface or a separate treasury management system.
Why use this instead of exporting bank statements manually?
Manual exports go stale the moment you download them, and you're re-uploading the same data every week. The Plaid MCP keeps a live connection, so your AI agents always see current balances and recent transactions without you touching a CSV. It also normalises data across different banks into a consistent format.
Who on the team should connect the Plaid MCP?
Whoever has login credentials for the financial accounts you want to analyse. Typically that's your finance lead or a founder. Once connected, any Switchy team member with access to the relevant workspace can query the data through AI agents. The connection itself doesn't count against user seats.