01
Connect accounts to Warm
Link checking, credit cards, loans, investments, and other supported accounts so Warm can aggregate everything into one place.
Warm Pro + MCP
Connect once. Reuse everywhere. The financial data layer for AI.
Warm syncs your accounts into one centralized transaction feed. After the sync completes, run npx @warmio/mcp and supported MCP clients can immediately work against the same read-only data.
One API key Read-only access Centralized sync Reuse across clients
Workflow
The setup is simple because the hard part is centralized.
Warm handles the account aggregation and financial normalization once. The AI layer stays lightweight on top.
02
Let the sync settle
After the initial sync finishes, Warm has a centralized, normalized transaction feed ready for analysis.
03
Run one command
Generate an API key in Warm, then run npx @warmio/mcp to configure supported MCP clients.
04
Ask, build, compare
Use the same read-only feed for prompts, dashboards, agents, and experiments across whichever client you want to try next.
Built-in Tools
Pro
Ask AI anythingabout your finances.
terminal
calcifer@mbp~%npx @warmio/mcp
Claude Codev1.0.49
✓Loaded warm tools (3 tools)
>How much did I spend on restaurants last month?
claude
Here's your restaurant spending last month:
Date Merchant Amount
──────── ──────────────────── ────────
Jan 02 Tartine Bakery $22.50
Jan 08 Delfina $94.50
Jan 14 Nopa $112.40
Jan 19 Souvla $38.20
Jan 26 Foreign Cinema $115.40
──────── ──────────────────── ────────
Total (5 transactions) $383.00
↑ 8% from December ($354.60)
Learn more about how Warm is AI friendly. Learn more →
Why now
Models will keep changing. Your financial context should not.
Warm is the picks-and-shovels layer: the reusable financial substrate you keep while clients, models, and prompt styles keep moving.
Stable underneath the model churn
New AI tools show up every week. Warm stays the durable layer below them so your finance context does not have to be rebuilt each time.
One source of financial truth
Instead of scattered exports, copied spreadsheets, or per-tool reconnect flows, Warm keeps one centralized feed ready for reuse.
Built for prompts and products
Use the same data to ask plain-English questions, generate a dashboard, compare model output, or power an internal agent workflow.
Supported clients
Plug the same feed into the tools you already test.
Claude Code Cursor Codex CLI Gemini CLI Windsurf Claude Desktop OpenCode Antigravity
Exposed read-only
The layer already contains the primitives you want to query.
Accounts and balances Transactions Recurring charges Net worth snapshots Budgets Goals Financial health Key verification
Security posture
Warm is useful as an AI bridge because it stays narrow.
Read-only by design
Warm can expose financial context to AI clients without giving them the ability to move money or edit your accounts.
Scoped to your data
Your API key only reaches your Warm data, so each MCP client sees your connected accounts and nothing beyond them.
Revocable in one click
Delete or rotate the key from Warm settings whenever you want to cut off a client or start fresh.
FAQ
Questions teams and solo builders will ask.
What does "financial data layer for AI" mean here?
Warm handles the hard part once: connecting accounts, syncing transactions, and organizing the data into a centralized financial feed. MCP-compatible AI clients then connect to that shared read-only layer instead of you rebuilding the setup for every new tool.
What is the install command?
The supported install flow is "npx @warmio/mcp". It prompts for your Warm API key and configures supported MCP clients automatically.
Which clients work today?
Warm currently supports Claude Code, Cursor, Codex CLI, Gemini CLI, Windsurf, Claude Desktop, OpenCode, and Antigravity, plus other MCP-compatible clients when configured appropriately.
What can an AI client access?
Warm exposes read-only tools around accounts, transactions, recurring items, snapshots, budgets, goals, health, and key verification. That gives the client enough context to answer questions, compare periods, and build on your financial data.
Is this only for chat prompts?
No. Prompting is the easiest starting point, but the same layer is useful for dashboards, personal agents, code-driven analysis, and experimentation across multiple AI clients.
Is Warm Pro required?
Yes. MCP and API access are part of Warm Pro, which is the plan intended for AI-ready finance workflows.
Final step
Keep Warm underneath, swap AI tools above it.
If the opportunity is to sell the picks and shovels for financial AI, Warm should be the durable layer you connect once and keep using everywhere.