AI Tools

Connect any MCP-compatible AI tool to your Chartcastr account in three steps. Pick your tool, run one command (or paste one block of JSON), restart, done.

Looking for the why instead of the how? See AI Tools — Concepts for the context-layer and action-agent patterns.

Step 1 — Get an API key

In the Chartcastr dashboard, open Settings → API Keys and create a new key. Copy it — you’ll paste it in Step 2.

Treat this key like a password. Anyone with it can read your sources, connections, and pulse analysis.

Step 2 — Install the MCP server in your tool

Pick your tool and run the matching command.

Claude Code
Claude Desktop
Cursor
ChatGPT Desktop
Codex

One line in your terminal:

claude mcp add --transport http chartcastr \
  https://api.chartcastr.com/api/mcp \
  --header "X-API-Key: YOUR_API_KEY"

That’s it — Claude Code persists this in your user config. Skip to Step 3.Prefer JSON? See the Claude Code page for the .mcp.json form.

Open the config file for your platform:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Add (or merge) this block, replacing YOUR_API_KEY:

{
  "mcpServers": {
    "chartcastr": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://api.chartcastr.com/api/mcp",
        "--header",
        "X-API-Key:YOUR_API_KEY"
      ]
    }
  }
}

Quit and reopen Claude. Full details on the Claude Desktop page.

Create or edit .cursor/mcp.json in your project (or ~/.cursor/mcp.json for global):

{
  "mcpServers": {
    "chartcastr": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://api.chartcastr.com/api/mcp",
        "--header",
        "X-API-Key:YOUR_API_KEY"
      ]
    }
  }
}

Open Cursor Settings → MCP and toggle chartcastr on. Switch to Agent mode to use it. Full details on the Cursor page.

Open ChatGPT Settings → MCP Servers → Add server:

Name: chartcastr
Command: npx
Args: -y mcp-remote https://api.chartcastr.com/api/mcp --header X-API-Key:YOUR_API_KEY

Restart ChatGPT. For GPT Actions (REST, no MCP) and full details, see the ChatGPT page.

Edit ~/.codex/config.yaml:

mcpServers:
  chartcastr:
    command: npx
    args:
      - "-y"
      - "mcp-remote"
      - "https://api.chartcastr.com/api/mcp"
      - "--header"
      - "X-API-Key:YOUR_API_KEY"

Run codex. Full details on the Codex page.

Step 3 — Verify it works

Ask your tool one of these:

“List my Chartcastr sources.”
“What did the latest pulse say about my HubSpot connection?”
“Are any of my connections inactive?”

If you see real data come back, you’re done. If not, jump to Troubleshooting. In Claude Code or Claude Desktop, you can also run /mcp to see the active server list — chartcastr should appear.

Available tools

These are the tools your AI agent can call once connected:

Tool	What it returns
`list_sources`	Every connected data source with status and connection count
`list_destinations`	Every configured destination (Slack workspace, email, etc.)
`list_connections`	Every active chart delivery — source, destination, cadence
`get_latest_pulse`	Full AI narrative analysis from the most recent delivery
`verify_connection`	Live confirmation a specific connection is active
`open_chartcastr`	Deep link to open the dashboard at a specific page

Local development

Running Chartcastr locally with bun up? Swap the URL to http://localhost:8109/api/mcp and use a key generated against your local server.

claude mcp add --transport http chartcastr-local \
  http://localhost:8109/api/mcp \
  --header "X-API-Key: YOUR_LOCAL_KEY"

Troubleshooting

401 Unauthorized

Your key is wrong, expired, or revoked. Regenerate at Settings → API Keys and re-run the install command.

Tool doesn't see chartcastr after install

Most MCP clients only load servers at startup. Quit fully and relaunch — a window reload isn’t enough. In Claude Code, run /mcp to force a server-list refresh.

`npx mcp-remote` hangs or errors

First run downloads the mcp-remote package — give it ~30 seconds. If it persists, run npx -y mcp-remote@latest once manually to prime the cache.

Tool returns 'not currently available'

A specific tool may be feature-flagged off during rollout. Check status.chartcastr.com or contact support.

Getting Started

For Agencies

Sources

Destinations

Analysis

Stats

API

Streams

Admin

AI Tools — Setup

Step 1 — Get an API key

Step 2 — Install the MCP server in your tool

Step 3 — Verify it works

Available tools

Local development

Troubleshooting

Getting Started

For Agencies

Sources

Destinations

Analysis

Stats

AI Tools

API

Streams

Admin

Documentation Index

​Step 1 — Get an API key

​Step 2 — Install the MCP server in your tool

​Step 3 — Verify it works

​Available tools

​Local development

​Troubleshooting

Step 1 — Get an API key

Step 2 — Install the MCP server in your tool

Step 3 — Verify it works

Available tools

Local development

Troubleshooting