> ## Documentation Index
> Fetch the complete documentation index at: https://docs.myfundingmachine.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Connect Your AI to Funding Machine
> Give Claude, ChatGPT, or your own AI agents secure, read-only access to your Funding Machine data using the built-in MCP server.
# Connect Your AI to Funding Machine
Funding Machine ships with a built-in **MCP server** (Model Context Protocol) — the open standard AI assistants use to talk to external systems. Connect Claude, ChatGPT, or any MCP-capable agent and it can look up your applicants' credit summaries, funding plans, and deal stages directly, without copy-pasting.
Access is **read-only** and scoped to your account. Your AI can look up data, but it cannot change anything, submit deals, or touch billing.
## What your AI can do
Once connected, your assistant has these read-only tools:
| Tool | What it returns |
| ----------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Search contacts** | Find CRM contacts by name, email, or phone, and get the contact IDs the other tools use |
| **Credit summary** | Latest credit score, DTI, underwriting decision, estimated funding ranges, and credit-repair flags for an applicant, with no raw report data or PII |
| **Underwriting result** | The full stored underwriting outcome: decisions, status codes, funding ranges, workflow path, and manual-review flags. Never re-runs underwriting |
| **Funding plans** | The ordered funding plan: recommended lenders and cards, expected amounts, approval probability, and next steps |
| **Deal stage** | Where an applicant sits in your pipeline right now, including stage, status, value, and last activity |
| **Applicant pipeline** | The contacts your team is tracking, most recently updated first, with follow-up status and upcoming appointments |
| **Dashboard stats** | Aggregate numbers for your account over a recent window: report volume, average score, decision breakdown, and funding totals |
Example prompts once connected:
* *"Find the contact named Maria Lopez and give me her credit summary."*
* *"Summarize the funding plan for this applicant and draft a follow-up email."*
* *"Which applicants need follow-up this week, and who has an appointment set?"*
* *"How did our funding volume trend over the last 90 days?"*
## Step 1 — Create an API key
1. In Funding Machine, go to **Settings → Integrations**
2. Under **AI Integration (MCP)**, click **Create key**
3. Name the key after where you'll use it (e.g. "Claude Desktop — ops team")
4. **Copy the key immediately.** For security, the full key is shown only once. If you lose it, revoke it and create a new one.
Keys look like `fmk_...` and are scoped to your account only. You can create up to 10 active keys and revoke any of them at any time from the same screen.
Treat API keys like passwords. Anyone with the key can read your applicants' funding data. Never share keys in email or chat, and revoke a key right away if you think it leaked.
## Step 2 — Connect your assistant
Your MCP endpoint is:
```text theme={null}
https://marketplace-app.myfundingmachine.com/api/mcp
```
(The exact URL is also shown on the Settings → Integrations screen.)
### Claude (claude.ai and Claude Desktop)
1. Open **Settings → Connectors** (on claude.ai, under your profile; in Claude Desktop, under Settings)
2. Click **Add custom connector**
3. Enter the endpoint URL above
4. When prompted for authentication, choose **API key / Bearer token** and paste your `fmk_...` key
5. Start a new conversation — the Funding Machine tools appear in the tools menu
For Claude Desktop versions that use a JSON config file instead, add:
```json theme={null}
{
"mcpServers": {
"funding-machine": {
"type": "http",
"url": "https://marketplace-app.myfundingmachine.com/api/mcp",
"headers": {
"Authorization": "Bearer fmk_YOUR_KEY_HERE"
}
}
}
}
```
### ChatGPT
1. Open **Settings → Connectors** (requires a plan with custom connector support)
2. Click **Create** under custom connectors
3. Enter the endpoint URL and select **Bearer token** authentication
4. Paste your `fmk_...` key and save
5. Enable the connector in a conversation from the composer's tools menu
### Custom agents and frameworks
Any MCP client that supports the **Streamable HTTP** transport works. Pass the API key as a Bearer token:
```typescript theme={null}
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StreamableHTTPClientTransport } from "@modelcontextprotocol/sdk/client/streamableHttp.js";
const transport = new StreamableHTTPClientTransport(
new URL("https://marketplace-app.myfundingmachine.com/api/mcp"),
{
requestInit: {
headers: { Authorization: "Bearer fmk_YOUR_KEY_HERE" },
},
},
);
const client = new Client({ name: "my-agent", version: "1.0.0" });
await client.connect(transport);
const tools = await client.listTools();
```
## Managing access
Everything is controlled from **Settings → Integrations**:
* **See every key** — its name, when it was created, and when it was last used
* **Revoke instantly** — a revoked key stops working on the next request
* **Audit trail** — the "Recent agent activity" list shows every tool call your connected AIs have made, so you always know what they looked up
## Limits and security
* **Read-only**: no tool can create, modify, or delete anything
* **Account-scoped**: a key only ever sees data for the account it was created in
* **Rate-limited**: 60 requests per minute per key, so a runaway agent loop can't hammer your data
* **No raw PII**: credit tools return summaries and decisions, never full report contents
* **Hashed at rest**: we store only a one-way hash of your key — nobody, including support, can recover the full key after creation
Create one key per assistant or teammate instead of sharing a single key. When someone leaves or a tool is retired, revoke just that key without disrupting everything else.