plaid

What This Skill Does

The Plaid skill integrates the plaid-cli tool into your OpenClaw agent, providing a robust interface for interacting with the Plaid financial services platform. This skill enables the agent to securely link banking institutions, retrieve real-time account balances, and perform deep analysis on transaction histories. By leveraging this tool, your agent can effectively act as a personal financial assistant, capable of reconciling accounts, monitoring spending habits, and automating the retrieval of financial data across multiple linked accounts without the user needing to log into individual banking portals.

Installation

To install this skill, run the following command within your environment: clawhub install openclaw/skills/skills/bobdevibecoder/bobagent-plaid

Ensure that you have Go installed on your system. Once the package is installed, configure your environment variables (PLAID_CLIENT_ID, PLAID_SECRET, and PLAID_ENVIRONMENT) to authenticate with Plaid. For persistent configuration, you can utilize the ~/.plaid-cli/config.toml file. The data directory ~/.plaid-cli will be created automatically to store secure tokens and account aliases.

Use Cases

This skill is ideal for users who want to automate financial tracking. Use it to generate monthly spending reports by querying transactions within a specific date range, identifying specific recurring subscriptions, or checking real-time liquidity across different investment and checking accounts. It is particularly powerful for developers or power users who want to pipe financial data into existing workflows like jq or awk for custom reporting. The skill also allows for monitoring; by creating a simple cron job script, the agent can notify you when new transactions hit your account.

Example Prompts

1. "Show me the current balances for all my linked Chase and Amex accounts."
2. "Search for all transactions at Starbucks or Dunkin Donuts from last month and calculate the total spent."
3. "List all my linked financial institutions and tell me if any require re-authentication."

Tips & Limitations

Security is paramount: never ask the agent to print or log secrets like access tokens or client secrets. If the agent encounters an ITEM_LOGIN_REQUIRED error, simply run the link command for that specific item to refresh the credentials. Always use specific date ranges when querying transactions to optimize performance and keep the data output manageable. Note that direct access to the tokens command is discouraged, as it may expose sensitive credentials in plain text logs. When piping data to external tools, ensure that output formats are strictly set to JSON or CSV for maximum compatibility.