Agent Costcenter
Track and allocate LLM or agent spend across multiple agents so solo builders see who burned budget and why.
Overview
io.github.mdfifty50-boop/agent-costcenter is an MCP server for the Operate phase that tracks multi-agent billing and cost allocation through MCP.
What is this MCP server?
- Multi-agent billing and cost allocation tracking via MCP tools
- npm package agent-costcenter-mcp v0.1.3 with stdio local transport
- Designed for fleets of agents where per-agent attribution replaces one blended API bill
- Fits operate-phase iteration when you throttle or refactor expensive agent paths
- Source on GitHub at mdfifty50-boop/agent-costcenter-mcp for customization
- Version 0.1.3 in server manifest
- npm identifier agent-costcenter-mcp with stdio transport
- Repository https://github.com/mdfifty50-boop/agent-costcenter-mcp
What problem does it solve?
Several agents share one API key and you cannot tell which agent or workflow drove most of your LLM spend.
Who is it for?
Solo builders running multiple MCP agents who need attribution before scaling usage or adding paid features.
Skip if: Teams that only run a single cheap agent or already use enterprise FinOps dashboards with full cloud integration.
What do I get? / Deliverables
Cost allocation data is available via MCP so you can iterate agent designs with per-agent billing visibility instead of guessing from aggregate invoices.
- MCP-accessible cost and allocation records per agent context
- Clearer decisions when trimming or merging agent roles
- Local cost-center integration without a separate FinOps SaaS
Recommended MCP Servers
Journey fit
Cost and billing visibility matters most once agents run repeatedly in production-like loops, where spend discipline drives iteration decisions. operate → iterate covers tuning agent usage, budgets, and allocation as you refine what each agent is allowed to run.
How it compares
MCP cost-allocation tooling for agent fleets, not payment processing or generic cloud cost MCPs alone.
Common Questions / FAQ
Who is io.github.mdfifty50-boop/agent-costcenter for?
Builders operating multiple AI agents who need billing and allocation tracking exposed to their MCP-enabled coding assistant.
When should I use io.github.mdfifty50-boop/agent-costcenter?
Use it during operate and grow reviews when token costs climb and you must allocate spend per agent before changing prompts or architecture.
How do I add io.github.mdfifty50-boop/agent-costcenter to my agent?
Install agent-costcenter-mcp from npm, configure it as a stdio MCP server in your client, and enable the cost tracking tools after restart.