Connect an agent over MCP
Point Claude, ChatGPT, Cursor, or anything else that speaks MCP at Era. It signs in with OAuth 2.1, you approve the connection on your own screen, and there's nothing Era-specific to install.
This is the widest way in: accounts, transactions, spending breakdowns, cash-flow forecasts, and a memory your agents share. With your approval it writes too — recategorizing, tagging, and setting up rules that keep running.
Open protocol, not a proprietary lock-in
Built on the Model Context Protocol (MCP), the open standard for agent connectivity — any compliant client plugs in without custom integration work.
Connected in minutes
Add the server URL as a Streamable HTTP MCP server in your client, approve the OAuth scopes, and ask a read-only question first — that's the whole setup.
One schema across every bank
Every connection lands in the same schema, whatever the bank — typed fields, never a statement export. Most run through regulated aggregators and carry balances and transactions; most of those transactions arrive categorized. Where a data source sends no category, your rules and your agent set it.
Permissioned by design
Every tool call is scoped to what the user has explicitly allowed, protected by bank-level encryption — and access can be revoked at any time.
Memory that travels between agents
Preferences, facts, and context live with the account, not with any single client — every connected agent reads and writes to the same store.
Works with the tools you already use
Claude, ChatGPT, Cursor, VS Code, or a generic Streamable HTTP client — if it speaks MCP, it can connect to era.
For example
“What did I spend on groceries last month, and how does that compare to the month before?”
“Tag every Uber charge as commuting, and make a rule so new ones get tagged too.”