nutrition-mcp: Instructions file for Claude Code

CLAUDE.md

nutrition-mcp CLAUDE.md is an instructions file for Claude Code from akutishevsky/nutrition-mcp. It costs 29,623 tokens per session, scanned A, original, MIT.

Repository instructions for nutrition-mcp, a server that lets compatible AI tools use nutrition-related functions. They explain the project’s Bun and TypeScript setup, its message-protocol endpoints, and how to test, deploy, and publish it.

In plain words
What is it for?
They guide work on the server entry point and protocol endpoint, including schemas, tests, deployment, and registry publishing. MCP means a standard way for AI tools to connect to external functions and data.
Why use it?
They give an AI coding assistant the project-specific rules it needs before changing or releasing the server. This reduces mistakes caused by using outdated library patterns or incorrect commands.

Instructions file for Claude Code

Written for Claude Code: the file is CLAUDE.md. Also seen: mentions CLAUDE.md; mentions subagents; names the TodoWrite tool.

This is akutishevsky/nutrition-mcp's own configuration. It tells Claude Code how to work on nutrition-mcp itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything nutrition-mcp configures →

Reuse

Borrowing it

Nothing to install: this file belongs to akutishevsky/nutrition-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/akutishevsky/nutrition-mcp/main/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/akutishevsky/nutrition-mcp

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for nutrition-mcp CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/akutishevsky/nutrition-mcp/claude-md/github.svg)](https://agentmods.dev/instructions/akutishevsky/nutrition-mcp/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/akutishevsky/nutrition-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/akutishevsky/nutrition-mcp/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for nutrition-mcp CLAUDE.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/akutishevsky/nutrition-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/akutishevsky/nutrition-mcp/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 29,623 This file is loaded in full into every session.
When invoked 29,623 The same file — it is already loaded in full.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.29623 $0.29623
Opus 5.5 $0.11849 $0.11849
Sonnet 5.5 $0.05925 $0.05925
Haiku 4.5 $0.02962 $0.02962

Measured 3d ago against content hash 7e01d46afcb7, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

nutrition-mcp CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 424 lines

How it starts

The opening of the file, as written. The whole thing — 424 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

nutrition-mcp is a Model Context Protocol (MCP) server for nutrition-related functionality, built with Bun and TypeScript. Entry point is src/index.ts. It runs on MCP TypeScript SDK v2 (@modelcontextprotocol/server at runtime, @modelcontextprotocol/client only in tests) with Zod 4 as an explicit dependency; @modelcontextprotocol/sdk v1 is gone, so import McpServer, InMemoryTransport, createMcpHandler from @modelcontextprotocol/server. Every inputSchema / outputSchema here is a schema object (z.object({…})), never a raw shape: v2 still auto-wraps raw Zod 4 shapes via a deprecated overload, but the explicit form is the supported path, and it lets the exported *_OUTPUT_SCHEMA constants be .parse()d directly in tests. /mcp is a dual-era endpoint built on createMcpHandler (src/mcp.ts): it serves the 2026-07-28 revision (per-request _meta envelope, server/discover, no protocol-level sessions, Mcp-Method / Mcp-Name routing headers validated against the body) and, through the default legacy: "stateless" fallback, still answers 2025-11-25 clients with the same per-request initialize idiom the old hand-rolled transport used — one server factory backs both eras. The factory gets no Hono context: the authenticated user travels in the authInfo.extra.userId pass-through from handleMcp, and the public origin (for the icon URL) is read from ctx.requestInfo. The "/mcp over HTTP" section of src/mcp.test.ts drives both eras in-process by pointing a v2 Client's fetch at the Hono route — there is no in-memory transport for the modern era, so that is the only way to test it. It lives inside mcp.test.ts deliberately: mock.module is process-wide, and a separate file with its own mock/restore of ./supabase.js broke middleware.test.ts's mock on Linux CI (not reproducible on macOS). Two rules follow: stub ./supabase.js from an existing mock window rather than opening another, and restore from a snapshot taken before mock.module — Bun patches the namespace in place, so restoring from the live import is a no-op. listChanged is advertised false on purpose: the endpoint is stateless and refuses GET, so no channel exists to deliver a list-changed notification. Two v2 wire facts worth knowing: advertised schemas are JSON Schema 2020-12 (v1 stamped draft-07), and a POST whose Content-Type is not application/json is answered 415 before the body is read. Server version must be updated in three places: package.json, src/mcp.ts (McpServer constructor), and server.json. The server icon is at public/favicon.ico. Tool call analytics (duration, success/failure, error category) are tracked via src/analytics.ts and persisted to a tool_analytics Supabase table. Each row also carries protocol_era and client_name, because the runtime access log is a ring buffer holding well under an hour of traffic and cannot answer whether anyone still uses the 2025-era leg; retiring it is gated on select count(distinct user_id) from tool_analytics where protocol_era = 'legacy' and invoked_at > now() - interval '30 days' reaching zero — distinct users, over a window long enough to include people who log weekly. client_name is present on every modern call but only on a legacy initialize, since the stateless legacy leg builds a fresh server per request and only that one carries clientInfo. Runtime log lines (console.*) never carry a user id, email or per-user path; that is a privacy-policy promise, pinned by src/log-privacy.test.ts, and analyticsLogLine is the only analytics log formatter. Per-user questions go to tool_analytics.

Read the full file on GitHub · 424 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago Changed · +1 lines · +788 tokens per session 7e01d46afcb7
  2. 18d ago Changed · +13 lines · +1,398 tokens per session 7f2dfa67f4d2
  3. 20d ago Changed · +6 tokens per session de09f0c90b7d
  4. 24d ago Changed · +24 lines · +5,400 tokens per session 491ab2f787a3
  5. 26d ago Changed · +1 lines · +494 tokens per session c3316cb57114
  6. 27d ago Changed · +11 lines · +820 tokens per session c51194ae6684
  7. 28d ago Changed · +2 lines · +451 tokens per session db441d2de8ae
  8. 29d ago Changed · +2 lines · +913 tokens per session e1315094b868
  9. 1mo ago Changed · +2 lines · +412 tokens per session d2ee108d4ea7
  10. 1mo ago First seen · 368 lines · 18,941 tokens per session scan A 5540b01b575b

Subscribe to this mod's changes

nutrition-mcp CLAUDE.md is an instructions file published in the GitHub repository akutishevsky/nutrition-mcp (83 stars, last pushed 2d ago), licensed MIT. It adds 29,623 tokens to every session, about $0.1185 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other instructions, from other repositories