Documentation Index
Fetch the complete documentation index at: https://grantmaster.dev/llms.txt
Use this file to discover all available pages before exploring further.
Integration Guide: MCP Server
Overview
GrantMaster includes a built-in Model Context Protocol (MCP) server that allows AI assistants like Claude to interact directly with your grant management data. The MCP server exposes 35+ tools and 5 resources for reading and managing projects, grants, expenses, compliance, and more.
Where: Navigate to Integrations > MCP Server in the sidebar
What It Does
| Capability | Description |
|---|
| Project Data | Read project details, milestones, budgets, and team assignments |
| Grant Management | Query grant applications, pipeline status, and active awards |
| Expense Tracking | View and categorize expenses, check budget utilization |
| Compliance | Check compliance status, review audit logs, and policy adherence |
| Document Access | Search and retrieve uploaded documents and reports |
| AI Agent Execution | Trigger and monitor agent runs through external AI clients |
Prerequisites
- Role: Administrator or user with
MANAGE_INTEGRATIONS permission
- Subscription: Growth Tier or higher (MCP server is not available on Free or Starter Tier)
- API Key: Required for authentication — generated from the MCP Server configuration page
Setup
Step 1: Navigate to MCP Server Settings
Go to Integrations in the main sidebar, then click Configure Server on the MCP Server card, or navigate directly to /integrations/mcp.
Step 2: Copy Your Endpoint URL
The MCP endpoint URL is displayed at the top of the page. It follows the format:
https://<region>-<project-id>.cloudfunctions.net/api/mcp
Step 3: Generate an API Key
- In the API Keys section, click New Key
- Give the key a descriptive name (e.g., “Claude Desktop — Work Laptop”)
- Copy the generated key immediately — it will not be shown again
- The key appears in the active keys table where you can revoke it later
Step 4: Connect Your AI Client
Use the connection guides on the MCP configuration page. Pre-formatted configurations are provided for:
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"grantmaster": {
"url": "<YOUR_ENDPOINT_URL>",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}
Claude Code
Add to your project’s .mcp.json:
{
"mcpServers": {
"grantmaster": {
"url": "<YOUR_ENDPOINT_URL>",
"headers": {
"Authorization": "Bearer <YOUR_API_KEY>"
}
}
}
}
Generic MCP Client
Use the HTTP transport with:
- URL: Your MCP endpoint URL
- Authentication:
Authorization: Bearer <YOUR_API_KEY> header
Usage
Once connected, your AI client can:
- Ask questions about your grant data: “What grants are due for reporting this month?”
- Run queries across projects: “Show me all projects over budget by more than 10%”
- Trigger actions like generating reports or categorizing expenses
- Access documents uploaded to the platform for analysis
All requests are scoped to your organization’s data and respect your RBAC permissions.
API Key Management
| Action | How |
|---|
| Create | Click “New Key” in the API Keys section |
| View | Active keys are listed with name, creation date, and last used timestamp |
| Revoke | Click the revoke button next to any key — takes effect immediately |
Security Best Practices
- Generate separate keys for each client or device
- Revoke keys you no longer use
- Do not share API keys in public repositories or chat messages
- Keys inherit your user permissions — a key created by an Admin has Admin-level access
Troubleshooting
| Issue | Solution |
|---|
| ”Unauthorized” error | Verify your API key is correct and not revoked |
| ”Forbidden” error | Check that your user role has the required permissions for the requested action |
| Connection timeout | Verify the endpoint URL is correct; check your network/firewall settings |
| No tools available | Ensure your subscription tier includes MCP access (Growth or higher) |
