mcp-gw: Instructions file for Claude Code

CLAUDE.md

mcp-gw CLAUDE.md is an instructions file for Claude Code from gpimthong/mcp-gw. It costs 1,503 tokens per session, scanned A, original, MIT.

Repository instructions for working on mcp-gw, a service that connects one client to multiple tool servers and a built-in searchable knowledge base.

In plain words
What is it for?
Installing, developing, building, running, and deploying the mcp-gw project, including its retrieval-based knowledge base.
Why use it?
It gives the coding agent the project’s architecture, commands, and development conventions in one place.

Instructions file for Claude Code

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

This is gpimthong/mcp-gw's own configuration. It tells Claude Code how to work on mcp-gw 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-gw configures →

Reuse

Borrowing it

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

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-gw CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/gpimthong/mcp-gw/claude-md.svg)](https://agentmods.dev/instructions/gpimthong/mcp-gw/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/gpimthong/mcp-gw/claude-md"><img src="https://agentmods.dev/badge/instructions/gpimthong/mcp-gw/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,503 This file is loaded in full into every session.
When invoked 1,503 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 $0.01503 $0.01503
Opus 5 $0.00751 $0.00751
Sonnet 5 $0.00301 $0.00301
Haiku 4.5 $0.00150 $0.00150

Measured 7d ago against content hash bc6ef53bdc9e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

mcp-gw 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 · 61 lines

How it starts

The opening of the file, as written. The whole thing — 61 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.

What this is

mcp-gw is an aggregating MCP (Model Context Protocol) gateway. It runs as a single HTTP endpoint that Claude Code (or any MCP client) connects to, and it fans out tool calls to multiple registered backend MCP servers plus a built-in RAG (retrieval-augmented generation) knowledge-base subsystem. Backend tools are namespaced as backend__tool_name to avoid collisions; RAG tools are namespaced rag__<kb_id> (per-KB search) and rag_admin__* (KB/doc management).

Commands

npm install
npm run dev      # tsx hot-reload against src/ (reads config from ./config/backends.json)
npm run build    # compile TypeScript -> dist/
npm start        # run compiled dist/index.js

There is no test suite and no lint script configured in package.json.

Docker (production and dev-in-container):

docker compose up -d                            # production image, port 3000
docker compose --profile dev up mcp-gateway-dev  # dev image: tsx watch, src/public bind-mounted, no rebuild needed on edits

config/backends.json is gitignored (holds backend auth headers/tokens) and must be created manually on a fresh clone (see README "First-time setup"). It is bind-mounted into the container and written to at runtime whenever backends are added/removed via the dashboard or REST API — treat it as generated state, not source.

Architecture

Everything is wired together in src/index.ts: a BackendRegistry (backend MCP client connections) and a RagManager (local knowledge bases) are both handed to a single ToolAggregator, which is the only thing that knows how to route a prefixed tool name to either a backend client or the RAG subsystem. The Express app (dashboard.ts) and the MCP HTTP routes (mcp-server.ts) both sit on top of the aggregator/registry/rag instances — there's no other communication path between them.

  • backend-registry.ts — owns MCP client connections to registered backends (Streamable HTTP or SSE transport). Maintains a 30s heartbeat per backend (_ping) that refreshes tools/status/latencyMs; a backend only appears in getAllConnected() once status === 'connected'. Reconnecting a backend fully tears down and recreates its client.
  • tool-aggregator.ts — the routing layer. listAllTools() merges namespaced backend tools with RAG tools for the MCP tools/list response. callTool(prefixedName, args) checks rag.isRagTool() first, otherwise splits on the first __ to get backendName/toolName and calls through registry.getClient(backendName). All backend tool calls are logged (start + duration + ok/error) via logger.
  • mcp-server.ts — implements the MCP Streamable HTTP transport by hand (session map keyed by mcp-session-id header, one Server+transport pair created per initialize call). If you need to change session lifecycle behavior, this is the only place it's handled.
  • config.ts — flat load/save of config/backends.json (path overridable via CONFIG_PATH). No validation beyond JSON shape; the dashboard/API are the enforcement point for well-formed BackendConfig.
  • logger.ts — in-memory circular buffer + SSE emitter, backs both the dashboard's live log view and /api/logs/stream.
  • dashboard.ts — Express app exposing the REST API (/api/health, /api/backends, /api/logs*, RAG doc/KB endpoints) and serving public/index.html (a no-build-step SPA).

Read the full file on GitHub · 61 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 · 61 lines · 1,503 tokens per session scan A bc6ef53bdc9e

Subscribe to this mod's changes

mcp-gw CLAUDE.md is an instructions file published in the GitHub repository gpimthong/mcp-gw (0 stars, last pushed 26d ago), licensed MIT. It adds 1,503 tokens to every session, about $0.0075 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

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

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

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

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

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