mcp-debug-claude-desktop: Instructions file for Claude Code

CLAUDE.md

mcp-debug-claude-desktop CLAUDE.md is an instructions file for Claude Code from rsascha/mcp-debug-claude-desktop. It costs 5,031 tokens per session, scanned A, original, MIT.

Repository instructions for a small test project that reproduces a bug in Claude Desktop’s MCP client, the part that connects the app to external tools. They record measured results and rule out tests that have already been disproved.

In plain words
What is it for?
Use them when investigating the multi-chat tool-call bug, preparing a reproducible GitHub project, or writing a bug report for Anthropic.
Why use it?
They prevent developers from repeating costly experiments or following explanations that evidence has already rejected. They keep work focused on finding where messages are lost.

Instructions file for Claude Code

Written for Claude Code: the file is CLAUDE.md. Also seen: mentions CLAUDE.md.

This is rsascha/mcp-debug-claude-desktop's own configuration. It tells Claude Code how to work on mcp-debug-claude-desktop 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 mcp-debug-claude-desktop configures →

Reuse

Borrowing it

Nothing to install: this file belongs to rsascha/mcp-debug-claude-desktop. 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/rsascha/mcp-debug-claude-desktop/main/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/rsascha/mcp-debug-claude-desktop

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 mcp-debug-claude-desktop CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/rsascha/mcp-debug-claude-desktop/claude-md.svg)](https://agentmods.dev/instructions/rsascha/mcp-debug-claude-desktop/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/rsascha/mcp-debug-claude-desktop/claude-md"><img src="https://agentmods.dev/badge/instructions/rsascha/mcp-debug-claude-desktop/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 5,031 This file is loaded in full into every session.
When invoked 5,031 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 $0.05031 $0.05031
Opus 5 $0.02516 $0.02516
Sonnet 5 $0.01006 $0.01006
Haiku 4.5 $0.00503 $0.00503

Measured 7d ago against content hash 37ccda08bb4a, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

mcp-debug-claude-desktop 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 7d 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 · 351 lines

How it starts

The opening of the file, as written. The whole thing — 351 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — mcp-debug-claude-desktop

Guidance for working in this repository. Read this fully before writing code. It contains measurements that are expensive to repeat and hypotheses that are already refuted — re-testing them is wasted work.

What this project is

A minimal, self-contained reproduction of a bug in Claude Desktop's MCP client: with several chats open, tool calls are lost — some are never sent to the server, and at least one response that the server demonstrably wrote to stdout never reached the client. In every case the user is told the server is unresponsive or has crashed, which it is not.

The deliverable is a public GitHub repository that Anthropic can clone and run in a few minutes, plus a bug report written from its output.

This repo contains no journal, no domain logic, no product. Its only job is to make one protocol-level question answerable: where does the message go?

Origin — what has already been measured

The bug was found on 2026-08-04 while operating an unrelated personal MCP server (ai-journal-mcp, a private journaling server). That server already carries structured instrumentation, which is how the following is known. Do not re-derive any of it.

The incident

Environment of that incident: Claude Desktop 1.24012.11 (macOS 26.5.2, arm64), Node v25.9.0, @modelcontextprotocol/sdk 1.29.0 — one version below the one this repository pins. The controlled runs of 2026-08-04 reproduced the same behaviour on 1.30.0.

Three searches were submitted in three chats, in this order: pidtest-delta, pidtest-echo, pidtest-foxtrot. The complete server log for that window of time — both server processes, every line:

{"ts":"13:02:43.044","pid":24774,"kind":"session","msg":"start","ppid":24769,"version":"0.1.55"}
{"ts":"13:02:43.347","pid":24801,"kind":"session","msg":"start","ppid":24796,"version":"0.1.55"}
{"ts":"13:03:59.407","pid":24774,"phase":"start","rpc_id":3,"tool":"search_journal","args":"query=\"pidtest-foxtrot\""}
{"ts":"13:04:00.797","pid":24774,"phase":"end","rpc_id":3,"status":"ok","ms":1391}
{"ts":"13:04:00.798","pid":24774,"phase":"sent","rpc_id":3,"ms":0,"bytes":2371,"blocked":false}
{"ts":"13:08:10.909","pid":24774,"phase":"start","rpc_id":4,"tool":"get_time"}
{"ts":"13:08:10.923","pid":24774,"phase":"end","rpc_id":4,"status":"ok","ms":14}
{"ts":"13:08:10.923","pid":24774,"phase":"sent","rpc_id":4,"ms":0,"bytes":160,"blocked":false}

Read the full file on GitHub · 351 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. 7d ago First seen · 351 lines · 5,031 tokens per session scan A 37ccda08bb4a

Subscribe to this mod's changes

mcp-debug-claude-desktop CLAUDE.md is an instructions file published in the GitHub repository rsascha/mcp-debug-claude-desktop (0 stars, last pushed 1mo ago), licensed MIT. It adds 5,031 tokens to every session, about $0.0252 per session on Opus 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 2026-08-31.

Related

Other instructions, from other repositories

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,469 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

spec-kit AGENTS.md

AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.

github/spec-kit · 7,104 tokens