Antenna MCP
Antenna provides a Model Context Protocol (MCP) server that lets AI assistants and agents query your engineering metrics directly. Use natural language prompts to explore contributor productivity, AI tool adoption, team trends, and more.
The MCP server exposes read-only tools scoped to your enterprise. All requests are authenticated with an Enterprise Access Token and return structured JSON responses.
Example Prompts
Once connected, you can ask your AI assistant natural language questions. The assistant will automatically select and call the appropriate MCP tools.
- “Who is spending the most on AI?”
- “Who used AI the most and least efficiently over the last 6 months?”
- “How can my team optimize AI spending?”
- “Whose PRs are taking the longest to be reviewed? Is there a correlation with PR size?”
- “Looking at the last 6 months, what are the biggest opportunities for improvement on my team?”
Prerequisites
- An Antenna enterprise account with the MCP feature enabled
- An Enterprise Access Token (generated in Settings > Access Tokens)
Generate an access token
- Sign in to app.antenna.dev as an enterprise admin.
- Go to your enterprise Settings > Access Tokens.
- Click Create token, give it a description (e.g., “MCP Server”), and copy the token.
The token is shown only once. Store it securely — you will need it for each MCP client configuration below.
Available Tools
The MCP server exposes the following read-only tools:
| Tool | Description |
|---|---|
contributor_metric_fields | Describe available metric fields, sort fields, date granularities, and result limits. |
enterprise_identity_groups | List identity groups (teams) available for filtering contributor metrics. |
enterprise_identity_search | Search enterprise identities by display name, username, email, or external account ID. |
contributor_metrics_list | Return contributor metrics for an enterprise over a bounded date range. |
contributor_metrics_trends | Return per-period contributor metrics for selected enterprise identities. |
Setup
The Antenna MCP server uses the Streamable HTTP transport. The endpoint is:
https://app.antenna.dev/api/mcp/enterpriseAll requests require an Authorization header with your enterprise access token:
Authorization: Bearer YOUR_ACCESS_TOKENJump to setup:
ChatGPT / Codex
OpenAI’s ChatGPT and Codex support MCP servers via their tool configuration.
Open ChatGPT settings
In ChatGPT (desktop or web), go to Settings > Tools & Integrations > Add MCP Server.
Configure the server
Enter the following:
- Name:
Antenna - Transport: Streamable HTTP
- URL:
https://app.antenna.dev/api/mcp/enterprise - Authentication: Bearer token
- Token: Your Enterprise Access Token
Save and start using
Once saved, ChatGPT can call Antenna tools in your conversations. Ask questions like “Who is spending the most on AI?” and it will use the MCP tools automatically.
For Codex (the CLI agent), add the server to your project configuration file at .codex/config.json:
{
"mcpServers": {
"antenna": {
"type": "streamable-http",
"url": "https://app.antenna.dev/api/mcp/enterprise",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}Claude Code (Desktop)
Claude Code supports MCP servers through its settings file.
Open Claude Code settings
Open your Claude Code settings file at ~/.claude/settings.json (or create it if it doesn’t exist).
Add the MCP server configuration
Add an entry under the mcpServers key:
{
"mcpServers": {
"antenna": {
"type": "streamable-http",
"url": "https://app.antenna.dev/api/mcp/enterprise",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}Restart Claude Code
Restart Claude Code (or open a new session) for the changes to take effect. You can verify the connection by asking Claude to list available Antenna tools.
For Claude Desktop (the chat app), add the server to your config at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"antenna": {
"type": "streamable-http",
"url": "https://app.antenna.dev/api/mcp/enterprise",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}Claude on AWS Bedrock
When using Claude via AWS Bedrock with MCP support, configure the Antenna MCP server in your Bedrock agent or application code.
Define the MCP server in your agent configuration
When creating or updating a Bedrock agent with MCP tool use, add the Antenna server to your tool configuration:
{
"mcpServers": {
"antenna": {
"url": "https://app.antenna.dev/api/mcp/enterprise",
"transport": "streamable-http",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}Configure in application code (Python example)
If you are invoking Claude on Bedrock programmatically with MCP tools, pass the server configuration when setting up your client:
mcp_servers = {
"antenna": {
"url": "https://app.antenna.dev/api/mcp/enterprise",
"transport": "streamable-http",
"headers": {
"Authorization": f"Bearer {ANTENNA_ACCESS_TOKEN}"
}
}
}Test the connection
Send a prompt that uses the Antenna tools (e.g., “List the identity groups in my enterprise”) to verify the integration is working.
AWS Bedrock MCP support may require specific agent configurations or SDK versions. Refer to the AWS Bedrock documentation for the latest MCP integration details.
Cursor
Cursor supports MCP servers through its settings.
Open Cursor settings
Go to Cursor Settings > MCP > Add new global MCP server.
Add the MCP server configuration
Select Type: Streamable HTTP and enter:
{
"mcpServers": {
"antenna": {
"url": "https://app.antenna.dev/api/mcp/enterprise",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}Verify the connection
After saving, you should see the Antenna tools listed under the MCP section. Use Agent mode to ask questions about your engineering metrics.
Troubleshooting
| Issue | Solution |
|---|---|
401 Unauthorized | Verify your access token is correct and has not been revoked. Regenerate it in Settings > Access Tokens if needed. |
403 Forbidden | The MCP feature may not be enabled for your enterprise. Contact your Antenna account representative. |
| Tools not appearing | Restart your AI client after updating the configuration. Ensure the URL and headers are correct. |
| Empty results | Check your date range and filters. The default range is the last 6 months. Use contributor_metric_fields to see available options. |