Connect an AI agent

Point Claude, or any Model Context Protocol client, at your own finances over a read-only connection you can revoke.

Ember can open your finances to an AI assistant: Claude, or any other app that speaks the Model Context Protocol. You can then ask about your accounts, spending, and net worth in plain language instead of building a report. The connection is read-only, and off until you turn it on. You create a token, paste it into your assistant, and revoke it whenever you like.

What your agent can see

Once connected, your assistant can look up three things:

Your accounts. Every active account with its current balance, institution, and type. Debts show as negative, and a joint account shows your share.
Your net worth. The current figure, split into assets and debts, and into cash, investments, and liabilities. It's the same number the Net Worth report shows.
Your transactions. Cash-ledger activity, newest first, with the payee, category, account, amount, memo, and cleared status. Your assistant can narrow to one account or a date range.

Everything is scoped to you. A token reads your data and nobody else's.

What you can do with it

Your agent reads your ledger and never writes to it, by design: analysis, reporting, and review need nothing more than read access, so the connection ships with no tools that could alter your data. It also reaches nothing beyond the three lookups above, so no login details, bank credentials, or institution connections are ever in scope.

Within that, the useful work is the kind that would otherwise mean building a report by hand.

Analysis. Ask about patterns rather than figures, and let your assistant do the arithmetic across months, categories, and accounts.

"What did I spend on restaurants last month, and how does that compare to the three months before?"
"Which subscriptions am I paying that I haven't touched this year?"
"Find every transaction over $500 since January and tell me which ones look unusual."

Reporting. Ask for the finished artifact rather than the raw numbers. Your assistant can write the monthly summary, build the category table, or draft the note you send to an accountant or a partner.

"Write me a one-page summary of July: what came in, what went out, and where it went."
"Build a table of every category over $200 this quarter, sorted by total."

Reconciliation. Paste or upload a bank or card statement into your assistant and have it cross-check that statement against what Ember shows as cleared for the same account and period. It can tell you which rows are missing, which amounts disagree, and which look like duplicates. You then apply the fixes in Ember, so the audit trail stays yours.

"Here's my July Amex statement. Compare it against my Ember card ledger and list anything that doesn't match."

Planning. Ask the questions that need the whole picture at once, across cash and investments.

"What's my net worth, and how much of it is actually liquid?"
"Which account is my grocery spending coming out of, and is that deliberate?"

Because your assistant reads the real ledger rather than a summary, it can follow up on its own answers instead of stopping at the first number.

Step 1: create a token

Go to Settings → AI agent access.
Under Create a token, give it a name that tells you which assistant it belongs to, like "Claude Desktop" or "work laptop". You'll be glad of the name later, when you're deciding which one to revoke.
Optionally set it to expire after 30 days, 90 days, or a year. No expiry is fine for a token you'll keep using.
Click Create token.

Copy the token now. It's shown once, and once only. Ember stores a fingerprint of it rather than the token itself, so it genuinely cannot show it to you a second time. If you lose it, revoke it and create another.

While you're on the page, copy the connection URL too. Your assistant needs both.

If you don't see AI agent access in Settings, the feature isn't switched on for your account yet.

Step 2: connect your assistant

Your token goes in an Authorization header, as Bearer followed by the token. How you enter that depends on the app.

Claude Code

One command, with your connection URL and token:

claude mcp add --transport http ember https://your-ember-host/mcp --header "Authorization: Bearer emb_mcp_your_token_here"

Claude web and desktop

Open Settings → Connectors → Add custom connector, enter the connection URL, then open the Request headers section and add a header named authorization with the value Bearer followed by your token, including the space after Bearer.

Request-header authentication is a beta feature that Anthropic is still rolling out, so the Request headers section may not appear in your account yet. If it doesn't, Claude Code above is the reliable route today.

Any other MCP client

Ember's endpoint is a standard remote MCP server over streamable HTTP. Point your client at the connection URL and have it send:

Authorization: Bearer emb_mcp_your_token_here

Clients vary in where that goes: a config file, a settings dialog, an environment variable. Any client that can attach a bearer token to a remote MCP server will work.

Managing your tokens

Settings → AI agent access lists every token you've created: its name, when you made it, when it was last used, and when it expires. The last-used date is worth a glance now and then. A token that's never been used is one you can safely revoke, and one being used when you aren't working is worth asking about.

Revoking is immediate and permanent. The agent holding that token loses access on its very next request, and the token can never be made to work again, so to reconnect that assistant you create a new one. Give each assistant its own token, so revoking one doesn't disconnect the rest.

Treat a token like a password: it reads your entire financial picture. Don't paste it into a shared config file, a chat message, or anywhere it'll end up in a repository.