Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/barkain/madrox/claude-mdgit clone --depth 1 https://github.com/barkain/madroxWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03714 | $0.03714 |
| Opus 5 | $0.01857 | $0.01857 |
| Sonnet 5 | $0.00743 | $0.00743 |
| Haiku 4.5 | $0.00371 | $0.00371 |
Grade B, and why
madrox CLAUDE.md scanned grade B with 2 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
# Example Codex CLI config (~/.codex/config.toml): Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
- Mock `subprocess.Popen` for CLI process simulation How it starts
The opening of the file, as written. The whole thing — 342 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.
Commands
Development Setup
# Install dependencies
uv sync --all-groups
# Activate virtual environment
source .venv/bin/activate
Plugin Mode (Claude Code)
Madrox can be installed as a Claude Code plugin. The plugin is configured via .mcp.json and managed by start_plugin.sh.
# Install from within Claude Code
/plugin install /path/to/madrox
How it works:
- Claude Code runs
start_plugin.shon startup - The script starts the HTTP backend (port 8001) and Next.js frontend dashboard (port 3002) in the background
- Waits for the backend health check to pass
- Runs a STDIO MCP proxy in the foreground (Claude Code connects to this)
- On exit, background processes are cleaned up automatically
- The STDIO server is a proxy — all tool calls are forwarded to the HTTP backend via HTTP
/madrox:dashboard skill: Opens the Madrox Monitor dashboard at http://localhost:3002 in the browser. Defined in skills/dashboard/SKILL.md.
Running the Server
The server supports two transport modes that are auto-detected:
HTTP/SSE Transport (Claude Code):
# Start HTTP server (default when running in terminal)
python run_orchestrator.py
# Server starts on http://localhost:8001
# Uses Server-Sent Events (SSE) for MCP protocol
STDIO Transport (Codex CLI):
# STDIO mode is auto-activated when stdin is piped
# Used by Codex CLI MCP client configuration
# Force STDIO mode via environment variable
MADROX_TRANSPORT=stdio python run_orchestrator.py
# Example Codex CLI config (~/.codex/config.toml):
# [mcp_servers.madrox]
# command = "python"
# args = ["/path/to/madrox/run_orchestrator.py"]
# env = { MADROX_TRANSPORT = "stdio" }
Quick Start with start.sh:
# Start both backend (port 8001) and frontend dashboard (port 3002)
./start.sh
# Start only the backend
./start.sh --be
# Start only the frontend
./start.sh --fe
Note: When backgrounding the backend, set MADROX_TRANSPORT=http to prevent STDIO auto-detection.
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.
- 3d ago First seen · 342 lines · 3,714 tokens per session scan B 454e72e0a644
madrox CLAUDE.md is an instructions file published in the GitHub repository barkain/madrox (5 stars, last pushed 1mo ago), licensed MIT. It adds 3,714 tokens to every session, about $0.0186 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
agent-framework copilot-instructions.md
Instructions for microsoft/agent-framework, covering github copilot instructions, repository structure and architectural decision records (adrs).
agent-framework python.instructions.md
Instructions for microsoft/agent-framework: See AGENTS.md for project structure and package documentation. Detailed conventions are in the agent skills under .github/skills/.
dark-factory AGENTS.md
Instructions for DUBSOpenHub/dark-factory, covering agents.md — working guide for ai agents, architecture, file ownership map, before you change anything and agent prompt rules.
paperclip-plugin-company-wizard CLAUDE.md
Instructions for Yesterday-AI/paperclip-plugin-company-wizard, covering claude.md, what this is, commands, architecture and source layout.
dark-factory copilot-instructions.md
Instructions for DUBSOpenHub/dark-factory, covering copilot instructions — dark factory, file map, non-negotiables, prohibited actions and pr requirements.
team-shinchan AGENTS.md
Instructions for seokan-jeong/team-shinchan, covering agents.md - team-shinchan agent map, layer architecture, agent registry, call flow and quick reference.