weckr — Claude Code MCP server: what it does, how to install
Queries AI expense figures, per-user margins, and feature-level costs directly inside conversation with your coding agent.
claude mcp add --env WECKR_API_KEY=<WECKR_API_KEY> --env WECKR_PROJECT_ID=<WECKR_PROJECT_ID> --env WECKR_BASE_URL=<WECKR_BASE_URL> --transport stdio mcp -- npx -y @weckr/mcpWhat weckr does
The server runs locally via npx and communicates over stdio with the Weckr HTTP API. It requires a WECKR_API_KEY with the wk_ prefix. At launch, the process queries /api/v1/me to resolve the project identifier. Subsequent requests are scoped to that project. The server sends no prompt text or completions.
Available tools inspect general performance with get_overview, list negative-margin users with get_users, and calculate feature cost shares with get_feature_breakdown. Other tools recommend cheaper models from the same provider or review pricing plans. Tool get_spending_cap_url provides a dashboard link without altering any limits. Passing an unmatched WECKR_PROJECT_ID triggers a 404 error from the API.
What it helps with
- identify users who consume more LLM spend than their subscription covers
- inspect which product features generate the largest AI bills
- receive cheaper model replacement suggestions within the same provider
- verify pricing plan viability against real model usage numbers
Example prompts for the agent
- Show me my most unprofitable accounts this month using Weckr.
- Which product features are driving our highest LLM expenses?
- Check our overall project margin with the Weckr tools.
How to install weckr
Platform CLI
claude mcp add --env WECKR_API_KEY=<WECKR_API_KEY> --env WECKR_PROJECT_ID=<WECKR_PROJECT_ID> --env WECKR_BASE_URL=<WECKR_BASE_URL> --transport stdio mcp -- npx -y @weckr/mcpReplace <…> with your own values
Platform CLI
codex mcp add mcp --env WECKR_API_KEY=<WECKR_API_KEY> --env WECKR_PROJECT_ID=<WECKR_PROJECT_ID> --env WECKR_BASE_URL=<WECKR_BASE_URL> -- npx -y @weckr/mcpReplace <…> with your own values
Config file
{
"mcpServers": {
"mcp": {
"command": "npx",
"args": [
"-y",
"@weckr/mcp"
],
"env": {
"WECKR_API_KEY": "<WECKR_API_KEY>",
"WECKR_PROJECT_ID": "<WECKR_PROJECT_ID>",
"WECKR_BASE_URL": "<WECKR_BASE_URL>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@weckr/mcp"
],
"env": {
"WECKR_API_KEY": "<WECKR_API_KEY>",
"WECKR_PROJECT_ID": "<WECKR_PROJECT_ID>",
"WECKR_BASE_URL": "<WECKR_BASE_URL>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e WECKR_API_KEY=<WECKR_API_KEY> -e WECKR_PROJECT_ID=<WECKR_PROJECT_ID> -e WECKR_BASE_URL=<WECKR_BASE_URL> mcp npx -y @weckr/mcpReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about weckr
How to install weckr in Claude Code?
Run claude mcp add weckr -e WECKR_API_KEY=<WECKR_API_KEY> -- npx -y @weckr/mcp in your terminal, replacing the placeholder with your project key.
What does the weckr MCP server do?
It connects your coding agent to Weckr analytics to track model expenses. The assistant can list unprofitable users, break down feature expenditures, and suggest cheaper models.
Does weckr transmit user prompts or completions?
No. The server only handles aggregated cost and usage metrics. Prompt content and model outputs never leave your environment.
