The ChartPull MCP server lets AI assistants like Claude Desktop, VS Code Copilot, and Cursor access your org chart data. Ask questions about your organisation without leaving your AI workflow.
The MCP server requires the Integrations add-on. During your 14-day trial, all workspaces have full access.
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools and data sources. Think of it like USB for AI — a single standard that lets any AI client plug into any data source.
When you connect the ChartPull MCP server to your AI assistant, the assistant gains the ability to search your employee directory, look up reporting chains, list departments, and even ask natural-language questions about your org structure — all within the same conversation where you are doing your other work.
PII stripping on all responses
Every response from the MCP server is automatically stripped of personally identifiable information like phone numbers and personal email addresses before being returned to the AI client. This ensures that sensitive employee data is never sent to third-party AI models.Generate an API key
Go to Admin › API Keys in ChartPull and click "Create API Key." Give it a descriptive name like "MCP Server." Copy the key (it starts with cpk_) — you will not be able to see it again.
Choose your AI client
The ChartPull MCP server works with any client that supports the Model Context Protocol. The most popular options are Claude Desktop, VS Code (with Copilot or Claude extension), and Cursor.
Add the configuration
Open your AI client’s MCP configuration file and add the ChartPull server entry. See the configuration examples below for the exact syntax for each client.
Restart your AI client
Most AI clients require a restart to pick up new MCP server configurations. After restarting, the ChartPull tools should appear in your assistant’s tool list.
Test with a query
Ask your AI assistant something like "Who are the engineering managers?" or "Show me the reporting chain for Sarah Chen." The assistant will use the ChartPull MCP tools to answer.
Add the ChartPull MCP server to your preferred AI client using the configuration below. Replace cpk_your_api_key_here with your actual API key.
Edit your Claude Desktop configuration file. On macOS, it is located at ~/Library/Application Support/Claude/claude_desktop_config.json. On Windows, it is at %APPDATA%\Claude\claude_desktop_config.json.
{
"mcpServers": {
"chartpull": {
"url": "https://chartpull.com/api/mcp",
"headers": {
"Authorization": "Bearer cpk_your_api_key_here"
}
}
}
}The MCP server exposes 8 tools that AI assistants can call. Each tool has a rate limit to ensure fair usage across all customers.
| Tool | What it does | Rate limit |
|---|---|---|
| search_employees | Search employees by name, title, or department | 30 / 60s |
| get_employee | Fetch a specific employee by ID | 60 / 60s |
| get_direct_reports | List all direct reports for a manager | 30 / 60s |
| get_manager_chain | Trace the full reporting chain up to the CEO | 30 / 60s |
| get_org_tree | Retrieve the full org tree structure | 10 / 60s |
| list_departments | List all departments with headcounts | 30 / 60s |
| search_natural_language | Ask a plain-English question (AI-powered) | 10 / 60s |
| get_org_insights | Get organisation health score and key metrics | 5 / 3600s |
get_org_insights is heavily rate-limited
Theget_org_insights tool is limited to 5 calls per hour because it triggers an AI-powered analysis of your entire organisation. For frequent lookups, use search_employees or list_departments instead.The MCP server authenticates using your ChartPull API key. The key uses the cpk_ prefix format and is passed as a Bearer token in the Authorization header. This is the same API key format used by the REST API, so if you already have an API key, you can reuse it.
Keep your API key secure
Your API key grants read access to your org chart data. Store it in your AI client’s configuration file, which is typically stored locally on your machine. Never share it in Slack channels, emails, or public repositories.Real-world scenario
DevOps Inc — 250 employees, tech company
Jake, a senior engineer at DevOps Inc, frequently needs to look up who owns which team, find the right manager to escalate to, or check reporting chains before writing design documents. Before ChartPull’s MCP server, he would context-switch out of his editor, open ChartPull in a browser tab, search for the person, and copy the information back.
After connecting the MCP server to his Cursor editor, Jake now asks questions like “Who are the engineering managers?” or “What’s the reporting chain from Alex Kim to the CEO?” directly in his AI assistant. The answer appears inline in seconds, and he never leaves his editor.
Jake estimates this saves him about 15 minutes per day — mostly from eliminating the context-switch tax of bouncing between tabs.