cd /news/ai-agents/the-mcp-context-tax-i-measured-what-… · home › topics › ai-agents › article
[ARTICLE · art-173392] src=dev.to ↗ pub= topic=ai-agents verified=true sentiment=· neutral

The MCP context tax: I measured what tool definitions cost your context window (21,328 MCP servers)

A developer measured the context-window cost of MCP tool definitions by calling tools/list on every public MCP server that answers without credentials, finding 13,716 responding servers out of 21,328. The median server's tool definitions consume roughly 4,800 tokens before any prompt is read, while the top 10% reach about 40,000 tokens — an 8× jump — with the longest lists topping out at 1,932 tools (minia2a), 1,040 (Novarithm Agent Hub) and 922 (DelegAIze). The author recommends trimming tool lists via the tools field and per-server toggles in clients like Cursor and Cline, noting that disabling 20 unused tools on a 30-tool server reclaims most of its cost.

recommendation

Audit your MCP servers' tool lists and disable tools you never call using the tools field or per-server toggles; the median server costs ~4.8k tokens per turn and the top 10% ~40k.

more in playbook: Controlling LLM cost
by read2 min views6 publishedNov 20, 2026

Originally published on Serverdex, where the numbers update live. Here's a cost most people never see: every MCP tool definition is loaded into your model's context before it does any work. Connect a server, and its tools' names, descriptions and full argument schemas are sent to the model on every single turn — before it reads your prompt.

Ten lean tools are cheap. One server with two hundred verbose tools is a standing bill you pay on every message. Nobody had measured that bill across the ecosystem, so I did.

I called tools/list on every public MCP server that answers — no API key, no OAuth — and measured the tool and token footprint. From the live run on 20 Nov 2026 (13,716 answering servers):

That bill isn't just tokens. It's:

Most answering servers sit in the 6–50 tool range, but the tail is heavy:

Tools exposed Servers
none listed 67
1–5 2,890
6–20 4,904
21–50 5,282
51+ 573

The jump from the median (~4.8k tokens) to the top 10% (~40k) is roughly 8×. A small number of servers carry most of the weight. The longest tool lists I found: minia2a (1,932 tools), Novarithm Agent Hub (1,040), DelegAIze (922).

Set against a 200K-token window (Claude's, used here only for scale): You rarely need every tool of every server in one task. From biggest effect to smallest:

tools field, Cursor's per-server enable switch, Cline's per-server toggles. Turning off the 20 tools you never call on a 30-tool server reclaims most of its bill. Serverdex calls every listed server with a streamable-HTTP remote about once a day with a credential-free initialize + tools/list, and records the tool list. Token cost is an estimate: characters of the raw definitions ÷ 4 (the floor — a real client frames each one with more). Percentiles are over answering servers.

If you maintain MCP servers: your tool list is a budget. Spend it like one.

── more in #ai-agents 4 stories · sorted by recency
── more on @serverdex 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/the-mcp-context-tax-…] indexed:0 read:2min 2026-11-20 · —