@metrxbot/mcp-server

v0.2.3

Published

2 months ago

Metrx MCP Server — AI agent cost intelligence tools for LLM agents. Track spend, optimize models, manage budgets, detect waste, and prove ROI.

Downloads

Metrx MCP Server

Your AI agents are wasting money. Metrx finds out how much, and fixes it.

The official MCP server for Metrx — the AI Agent Cost Intelligence Platform. Give any MCP-compatible agent (Claude, GPT, Gemini, Cursor, Windsurf) the ability to track its own costs, detect waste, optimize model selection, and prove ROI.

Why Metrx?

| Problem | What Metrx Does | |---------|-----------------| | No visibility into agent spend | Real-time cost dashboards per agent, model, and provider | | Overpaying for LLM calls | Provider arbitrage finds cheaper models for the same task | | Runaway costs | Budget enforcement with auto-pause when limits are hit | | Wasted tokens | Cost leak scanner detects retry storms, context bloat, model mismatch | | Can't prove AI ROI | Revenue attribution links agent actions to business outcomes |

Quick Start

Try it now — no signup required

npx @metrxbot/mcp-server --demo

This starts the server with sample data so you can explore all 23 tools instantly.

Connect your real data

Option A — Interactive login (recommended):

npx @metrxbot/mcp-server --auth

Opens your browser to get an API key, validates it, and saves it to ~/.metrxrc so you never need to set env vars.

Option B — Environment variable:

METRX_API_KEY=sk_live_your_key_here npx @metrxbot/mcp-server --test

Get your free API key at app.metrxbot.com/sign-up.

Add to your MCP client (Claude Desktop, Cursor, Windsurf)

If you used --auth, no env block is needed — the key is read from ~/.metrxrc automatically:

{
  "mcpServers": {
    "metrx": {
      "command": "npx",
      "args": ["@metrxbot/mcp-server"]
    }
  }
}

Or pass the key explicitly via environment:

{
  "mcpServers": {
    "metrx": {
      "command": "npx",
      "args": ["@metrxbot/mcp-server"],
      "env": {
        "METRX_API_KEY": "sk_live_your_key_here"
      }
    }
  }
}

Remote HTTP endpoint

For remote agents (no local install needed):

POST https://metrxbot.com/api/mcp
Authorization: Bearer sk_live_your_key_here
Content-Type: application/json

From npm

npm install @metrxbot/mcp-server

23 Tools Across 10 Domains

Dashboard (3 tools)

| Tool | Description | |------|-------------| | metrx_get_cost_summary | Comprehensive cost summary — total spend, call counts, error rates, and optimization opportunities | | metrx_list_agents | List all agents with status, category, cost metrics, and health indicators | | metrx_get_agent_detail | Detailed agent info including model, framework, cost breakdown, and performance history |

Optimization (4 tools)

| Tool | Description | |------|-------------| | metrx_get_optimization_recommendations | AI-powered cost optimization recommendations per agent or fleet-wide | | metrx_apply_optimization | One-click apply an optimization recommendation to an agent | | metrx_route_model | Model routing recommendation for a specific task based on complexity | | metrx_compare_models | Compare LLM model pricing and capabilities across providers |

Budgets (3 tools)

| Tool | Description | |------|-------------| | metrx_get_budget_status | Current status of all budget configurations with spend vs. limits | | metrx_set_budget | Create or update a budget with hard, soft, or monitor enforcement | | metrx_update_budget_mode | Change enforcement mode of an existing budget or pause/resume it |

Alerts (3 tools)

| Tool | Description | |------|-------------| | metrx_get_alerts | Active alerts and notifications for your agent fleet | | metrx_acknowledge_alert | Mark one or more alerts as read/acknowledged | | metrx_get_failure_predictions | Predictive failure analysis — identify agents likely to fail before it happens |

Experiments (3 tools)

| Tool | Description | |------|-------------| | metrx_create_model_experiment | Start an A/B test comparing two LLM models with traffic splitting | | metrx_get_experiment_results | Statistical significance, cost delta, and recommended action | | metrx_stop_experiment | Stop a running model routing experiment and lock in the winner |

Cost Leak Detector (1 tool)

| Tool | Description | |------|-------------| | metrx_run_cost_leak_scan | Comprehensive 7-check cost leak audit across your entire agent fleet |

Attribution (3 tools)

| Tool | Description | |------|-------------| | metrx_attribute_task | Link agent actions to business outcomes for ROI tracking | | metrx_get_task_roi | Calculate return on investment for an agent — costs vs. attributed outcomes | | metrx_get_attribution_report | Multi-source attribution report with confidence scores and top contributors |

Alert Configuration (1 tool)

| Tool | Description | |------|-------------| | metrx_configure_alert_threshold | Set cost or operational alert thresholds with email, webhook, or auto-pause |

ROI Audit (1 tool)

| Tool | Description | |------|-------------| | metrx_generate_roi_audit | Board-ready ROI audit report for your AI agent fleet |

Upgrade Justification (1 tool)

| Tool | Description | |------|-------------| | metrx_get_upgrade_justification | ROI report for tier upgrades based on current usage patterns |

Prompts

Pre-built prompt templates for common workflows:

| Prompt | Description | |--------|-------------| | analyze-costs | Comprehensive cost overview — spend breakdown, top agents, optimization opportunities | | find-savings | Discover optimization opportunities — model downgrades, caching, routing | | cost-leak-scan | Scan for waste patterns — retry storms, oversized contexts, model mismatch |

Examples

"How much am I spending?"

User: What was my AI cost this week?
→ metrx_get_cost_summary(period_days=7)

Total Spend: $234.56 | Calls: 2,450 | Error Rate: 0.2%
├── customer-support: $156.23 (1,800 calls)
└── code-generator: $78.33 (650 calls)

💡 Switch customer-support from GPT-4 to Claude Sonnet: Save $42/week

"Find me savings"

User: Am I overpaying for my agents?
→ metrx_compare_models(models=["gpt-4o", "claude-3-5-sonnet", "gemini-1.5-pro"])

Model Comparison (per 1M tokens):
├── gpt-4o: $2.50 in / $10.00 out
├── claude-3-5-sonnet: $3.00 in / $15.00 out
└── gemini-1.5-pro: $3.50 in / $10.50 out

"Test a cheaper model"

User: Test Claude 3.5 Sonnet against my GPT-4 setup
→ metrx_create_model_experiment(agent_id="agent_123",
    model_a="gpt-4o", model_b="claude-3-5-sonnet-20241022", traffic_split=10)

Experiment started: 90% GPT-4o, 10% Claude 3.5 Sonnet
Check back in 14 days for statistical significance.

Companion Tool: Cost Leak Detector

This repo also includes @metrxbot/cost-leak-detector — a free, offline CLI that scans your LLM API logs for wasted spend. No signup, no cloud, no data leaves your machine.

npx @metrxbot/cost-leak-detector demo

It runs 7 checks (idle agents, premium model overuse, missing caching, high error rates, context overflow, no budgets, arbitrage opportunities) and gives you a scored report in seconds. See the full docs.

Configuration

API Key (required)

The server looks for your API key in this order:

METRX_API_KEY environment variable
~/.metrxrc file (created by --auth)

Run npx @metrxbot/mcp-server --auth to save your key, or set the env var directly.

| Variable | Required | Description | |----------|----------|-------------| | METRX_API_KEY | Yes* | Your Metrx API key (get one free) | | METRX_API_URL | No | Override API base URL (default: https://metrxbot.com/api/v1) |

*Not required if you've run --auth — the key is read from ~/.metrxrc automatically.

CLI Flags

| Flag | Description | |------|-------------| | --demo | Start with sample data — no API key or signup needed | | --auth | Interactive login — opens browser, validates key, saves to ~/.metrxrc | | --test | Verify your API key and connection |

Rate Limiting

60 requests per minute per tool. For higher limits, contact [email protected].

Development

git clone https://github.com/metrxbots/mcp-server.git
cd mcp-server
npm install
npm run typecheck
npm test

Contributing

See CONTRIBUTING.md for guidelines.

A Note on Naming

The product is Metrx (metrxbot.com). The npm scope is @metrxbot and the Smithery listing is metrxbot/mcp-server. The GitHub organization is metrxbots (with an s) because metrxbot was already taken on GitHub. If you see metrxbot vs metrxbots across platforms, they're the same project — just a GitHub namespace constraint.

License

MIT — see LICENSE.

💬 Feedback

Did Metrx work for you? We'd love to hear it — good or bad.

GitHub Discussions: Start a thread — questions, ideas, what you're building
Bug reports: Open an issue
Quick feedback: Drop a comment on our Product Hunt listing

If you installed but hit a snag, tell us what happened — we read every report.