MCPSpend

Solo founder launch tomorrow MCPSpend is the cost-attribution layer for AI agent tooling. If you
use Cursor, Claude Desktop, Windsurf, or VS Code with MCP servers
(filesystem, GitHub, Playwright, Figma-remote, anything), MCPSpend
tracks every tool call: cost, latency, tokens, success/fail. Install is one command, auto-detects every MCP client on your machine: npx --yes @mcpspend/proxy@latest init --key mcps_live_xxx Free tier: 25k calls/month, no credit card Pro: $29/mo (or $290/yr 2 months free) Open source proxy on npm, IDE extension on Open VSX, MCP server on Smithery Why I built it: my Cursor + Claude bill ran 7x in three months and the
Anthropic console showed me a number but no breakdown. So I built the
observability layer I wished existed. What's already live: @mcpspend/proxy on npm (v0.5.0) @mcpspend/mcp-server (query your usage from inside the chat) mcpspend-vscode on Open VSX (Cursor + Windsurf compatible) Stripe billing, monthly/yearly, cancel anytime Dashboard at mcpspend.com Launching tomorrow at 12:01 PT. Would mean a lot if you set a
notification on the listing. Roast me hard in the comments solo
founder, any feedback helps. Link: https://www.producthunt.com/prod...

Hey everyone I'm Andrei, 33, solo founder out of Bucharest, Romania. After a decade
of building software first as an employee, last 2 years as an indie I'm finally on PH with a product I actually use every day myself. What I'm shipping: MCPSpend, real-time cost tracking for MCP (Model
Context Protocol) tool calls in Cursor, Claude Desktop, Windsurf, and
VS Code. Built it because my own AI tooling bill went from $40 to
$300+/month in three months and I had zero visibility into why. A few things about me: I built it solo, on a 10/mo VPS, in two weeks no team, no funding I use Cursor + Claude Code daily and ship 5+ commits/day on a good day Other side projects: an SaaS for builders (flowdeskone.ro), a hospitality booking site (anca-resort), all built from scratch Big believer in indie SaaS @levelsio and @marc_louvion are my playbooks Launching MCPSpend tomorrow on PH (12:01 PT). Would love to swap notes
with other solo founders here drop a hi 

I spent the last 30 days logging every MCP tool call I made across
Cursor, Claude Code, and Claude Desktop. Some surprises: 1. Read tool calls dominate by volume but tokens are tiny they're cheap. The cost spike comes from a handful of big calls (web search, image processing, large file reads). 2. ~14% of tool calls fail or timeout. They still cost something on the LLM side because the model already burned input tokens. 3. There's no native way in Cursor/Claude/Windsurf to see which project or which session caused the spend. You only see totals. 4. Per-agent attribution flips assumptions: "the Claude Code in my dev folder" vs. "the Cursor in my marketing folder" had a 7x cost delta for the same number of files touched. I turned the logger into a product (MCPSpend) and we're launching on
PH tomorrow. Free tier 25k calls/month if anyone wants to test:
npx --yes @mcpspend/proxy@latest init --key mcps_live_xxx Curious anyone else tracking their MCP spend manually? What's
your stack?