MCP tool discovery eats 10,000 tokens. I got it down to 350.

A developer created mcptoon, a CLI client that outputs Token-Optimized Object Notation (TOON) instead of JSON to reduce token overhead in MCP tool discovery and calls. The tool cuts tool discovery tokens from 2,034 to 62 for 96 tools, and in a real session reduced overhead from 47,200 to 8,100 tokens. It is designed to work with AI agents like Claude Code, Codex, and Cursor, and includes safety blocks for dangerous operations.

I use MCP servers with Claude Code every day. Last week I actually counted how many tokens get burned just on tool discovery. 5 servers, 96 tools total. The JSON listing: 2,034 tokens . On a 128K context window, that's before I've asked a single question. 20 tool calls later, each wrapped in {"content": {"type":"text","text":"..."} } — another 40,000 tokens of overhead. Brackets, quotes, commas, repeated {"type":"object","properties": declarations. I wrote mcptoon https://github.com/activeing123/mcptoon to deal with this. It's a CLI client that outputs TOON Token-Optimized Object Notation instead of JSON. The idea is dumb on purpose: the LLM doesn't need {"type":"object","properties": to understand what a tool does. It just needs the tool name and the relevant fields. pip install mcptoon Zero dependencies. 50KB. Python 3.10+. Windows, macOS, Linux. Here's tool discovery from 2 MCP servers: JSON 287 tokens — what every MCP client returns: {"name": "search web", "description": "Search the web for information", "inputSchema": {"type": "object", "properties": {"query": {"type": "string", "description": "Search query"}, "num results": {"type": "number", "default": 5}}, "required": "query" }}, {"name": "fetch url", "description": "Fetch content from a URL", "inputSchema": {"type": "object", "properties": {"url": {"type": "string"}}, "required": "url" }} TOON 5 tokens — what mcptoon returns: search web fetch url For tool discovery, the agent just needs to know what tools exist. Not the full schema every single time. When it does need the schema, TOON with full details is still 60% smaller: name:search web|description:Search the web|inputSchema:type:object|properties:query:type:string|description:Search query|num results:type:number|default:5|required:query|| name:fetch url|description:Fetch content from a URL|inputSchema:type:object|properties:url:type:string|required:url | Operation | JSON tokens | TOON tokens | Saved | |---|---|---|---| | Tool discovery 96 tools | 2,034 | 62 | 97% | | Tool result structured | 812 | 354 | 56% | | Tool result raw HTML | 1,023 | 912 | 11% | Real session: 5 servers, 20 tool calls. JSON overhead was 47,200 tokens. With mcptoon: 8,100 tokens. That's 39,100 tokens back. | JSON | TOON | What changed | |---|---|---| {"name":"search","count":3} | name:search\ | count:3 | 1, 2, 3 | 1 2 3 | Spaces instead of brackets+commas | true / false | T / F | 1 char vs 4-5 | null | ∅ | 1 symbol vs 4 | "line1\nline2" | line1↲line2 | ↲ instead of escape sequence | I tested this pretty thoroughly — Claude, GPT-4, and Gemini all parse TOON output correctly. The structure is recoverable from the compact form. What's not recoverable is the 1,900 tokens you spent on {"type":"object","properties": repeated 96 times. pip install mcptoon mcptoon init mcptoon add fetch --stdio npx -y @modelcontextprotocol/server-fetch mcptoon manifest --toon → fetch:fetch mcptoon call fetch fetch '{"url":"https://example.com"}' --toon Set MCPTOON AGENT TYPE=claude and it auto-selects --toon on every call. mcptoon is a CLI tool. If your agent runs shell commands, it can use it. No SDK, no plugin. | Agent | Setup | |---|---| | Claude Code | mcptoon in SKILL.md | | Codex | mcptoon in AGENTS.md | | Cursor | mcptoon in .cursorrules | | OpenCode | mcptoon in custom commands | | CatPaw | mcptoon in skill files | One config file ~/.mcptoon/config.json , every agent shares it. Add a server, all agents see it instantly. Dangerous operations get blocked by default: bash $ mcptoon call db delete table '{"name":"users"}' Error CONFIRMATION REQUIRED : Dangerous operation needs confirmation $ mcptoon call db delete table '{"name":"users"}' --destructive now it runs Patterns blocked: delete , drop , purge , wipe , kill , force=true , confirm=true . Pass --destructive to override. python from mcptoon.client import MCPClient from mcptoon.output import toon with MCPClient stdio= "npx", "-y", "@modelcontextprotocol/server-fetch" as c: tools = c.list tools print toon tools compact TOON result = c.call tool "fetch", {"url": "https://example.com"} print toon result src/mcptoon/ ├── cli.py CLI entry + arg parsing ├── client.py MCPClient — stdio + HTTP transport ├── router.py Tool routing, safety checks ├── output.py TOON / JSON / compact rendering ├── cache.py Schema cache 5-min TTL └── usage.py Local usage tracking 1,700 lines. Zero third-party imports. I have a personal grudge against dependencies for something this simple. | mcptoon | mcp-cli | raw MCP SDK | | |---|---|---|---| | Token savings | 97% manifest, 56% results | 0% | 0% | | All agents | yes | Claude only | varies | | Dependencies | 0 | 5-20 | 3-8 | | Install size | 50KB | 50MB+ | 10MB | | Platforms | Win+Mac+Linux | Linux/Mac | varies | GitHub: activeing123/mcptoon https://github.com/activeing123/mcptoon PyPI: mcptoon https://pypi.org/project/mcptoon/ License: Apache 2.0 | Dependencies: 0 | Tests: 98 passing in 0.09s Repo is here https://github.com/activeing123/mcptoon . Issues and feedback welcome — I'm especially curious if anyone has measured MCP token overhead in production and has different numbers than mine.