Last Updated: February 18, 2026
Trace supports the Model Context Protocol (MCP), an open standard that lets AI assistants read and write your Trace data. Connect Claude, ChatGPT, Gemini, or any MCP-compatible assistant to search entries, create new ones, organize streams, and more.
A Trace Pro subscription is required for integrations.
Select your AI assistant below for step-by-step setup instructions.
The simplest way to connect. No API key needed — you log in with your Trace account and Claude handles the rest.
Go to claude.ai and open Settings → Connectors. At the bottom of the page, click "Add custom connector".
Enter Trace as the name and paste the server URL:
https://trace-mcp.mindjig.com
Click Add.
Click Connect next to the Trace connector. A login page will open. Sign in with your Trace account (Google or email/password). Once authenticated, Claude automatically receives a secure token.
Trace will show a Configure button, confirming the connection is active. Trace tools will appear in your conversations.
Claude Desktop connects using an API key in a configuration file.
In the Trace mobile app, go to Settings → Integrations and tap Create. Choose a name and scope (Read Only or Full Access). Copy the key immediately — it won't be shown again.
Open your claude_desktop_config.json and add:
{
"mcpServers": {
"trace": {
"url": "https://trace-mcp.mindjig.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Replace YOUR_API_KEY with the key you copied.
Restart the app. Trace tools will be available in your conversations.
Same as above — create a key in the Trace app under Settings → Integrations.
Add to your project's .mcp.json:
{
"mcpServers": {
"trace": {
"type": "url",
"url": "https://trace-mcp.mindjig.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Trace tools will be available in your Claude Code sessions.
ChatGPT supports MCP servers through Developer Mode. It uses OAuth to connect — no API key needed.
Requirements: ChatGPT Pro, Plus, Business, Enterprise, or Education plan.
In ChatGPT, go to Settings → Connectors → Advanced and turn on Developer Mode.
In the Connectors settings, click Create. Enter a name (e.g. "Trace") and paste the server URL:
https://trace-mcp.mindjig.com
Click Create to add the connector.
ChatGPT will initiate an OAuth flow and redirect you to the Trace login page. Sign in with your Trace account (Google or email/password). Once authenticated, ChatGPT automatically attaches the token to all future requests.
Trace tools will appear in your ChatGPT conversations. You can review tool calls and their payloads before they execute.
Note: ChatGPT only supports OAuth — you cannot use an API key with ChatGPT. The server must be publicly accessible (no localhost).
Gemini CLI supports both OAuth (automatic) and API key authentication for MCP servers.
Gemini CLI can auto-discover OAuth endpoints. Add the server to your settings.json:
{
"mcpServers": {
"trace": {
"httpUrl": "https://trace-mcp.mindjig.com/mcp"
}
}
}
When Gemini CLI connects, it will detect the OAuth requirement, open a browser for login, and manage tokens automatically. Tokens are stored in ~/.gemini/mcp-oauth-tokens.json and refreshed when expired.
In the Trace mobile app, go to Settings → Integrations and tap Create. Choose a name and scope. Copy the key immediately.
Add to your Gemini CLI settings.json:
{
"mcpServers": {
"trace": {
"httpUrl": "https://trace-mcp.mindjig.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Replace YOUR_API_KEY with the key you copied.
You can also add a server via the command line:
gemini mcp add trace https://trace-mcp.mindjig.com/mcp \ --transport http \ --header "Authorization: Bearer YOUR_API_KEY"
Any MCP-compatible client can connect to Trace using an API key.
In the Trace mobile app, go to Settings → Integrations and tap Create. Choose a name and scope:
Copy the key immediately — it won't be shown again.
Point your MCP client at the server URL with a Bearer token header:
https://trace-mcp.mindjig.com/mcpAuthorization: Bearer YOUR_API_KEYIf your client supports OAuth, you can use https://trace-mcp.mindjig.com (without /mcp) and the OAuth flow will handle authentication automatically.
All connection methods provide access to the same set of tools:
| Tool | Description | Scope |
|---|---|---|
get_docs |
Get documentation about Trace's data model and best practices | Read |
list_streams |
List all streams (notebooks/collections) | Read |
get_stream |
Get details about a specific stream | Read |
list_entries |
List entries with optional filters | Read |
get_entry |
Get a single entry with full content | Read |
search_entries |
Full-text search across all entries | Read |
create_entry |
Create a new entry | Full |
update_entry |
Update an existing entry | Full |
delete_entry |
Delete an entry | Full |
get_attachment_url |
Get a signed URL for an attachment | Read |
get_attachment_data |
Get attachment image data (base64) | Read |
OAuth: "Login failed" or redirect error — Make sure you're logging in with the same account you use in the Trace app. Check that your Trace Pro subscription is active.
401 Unauthorized — Your API key is invalid or revoked. Create a new one in Settings → Integrations.
403 Forbidden — Your key doesn't have the required scope. For write operations (create, update, delete), you need a Full Access key.
Connection refused — Check that the server URL is exactly https://trace-mcp.mindjig.com/mcp (with /mcp) for API key connections, or https://trace-mcp.mindjig.com (without /mcp) for OAuth.
OAuth: "Token expired" — OAuth tokens last one year. If expired, remove the integration and re-add it to trigger a fresh login.
Contact us at support@mindjig.com.