What is agent-nero CLAUDE.md?

Project instructions for Claude Code working on Agent NERO, an MCP server that keeps named language-model agents alive with conversation history and shared memory. They describe its TypeScript structure, Node.js requirements, tools, and how to run it.

Use them when building, running, debugging, or modifying Agent NERO and its agent pool, memory bus, MCP tools, or TypeScript source.

They give the agent the project’s architecture and operating rules before it changes code, reducing confusion about where features belong.

agent-nero CLAUDE.md — Instructions file for coding agents by sanchez314c: Project instructions for Claude Code working on Agent NERO…

Install

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.

agentmods

npx agentmods add instructions/sanchez314c/agent-nero/claude-md

Clone the repo

git clone --depth 1 https://github.com/sanchez314c/agent-nero

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/sanchez314c/agent-nero/claude-md.svg)](https://agentmods.dev/instructions/sanchez314c/agent-nero/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/sanchez314c/agent-nero/claude-md"><img src="https://agentmods.dev/badge/instructions/sanchez314c/agent-nero/claude-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 822 This file is loaded in full into every session.

When invoked 822 The same file — it is already loaded in full.

Security scan C 1 finding. Scan, not verified.

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.

Model	Per session	Once invoked
Fable 5	$0.00822	$0.00822
Opus 5	$0.00411	$0.00411
Sonnet 5	$0.00164	$0.00164
Haiku 4.5	$0.00082	$0.00082

Measured 2d ago against content hash 00706fc5e215, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

agent-nero CLAUDE.md scanned grade C with 1 finding 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 2d 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.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- Destructive shell commands are blocked (rm -rf /, mkfs, dd if=, fork bombs).

CLAUDE.md · 79 lines

How it starts

The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — Agent NERO

Project-specific instructions for Claude Code sessions working on this codebase.

Project Overview

Agent NERO is a Model Context Protocol (MCP) server that provides persistent, named LLM agents to Claude Code. Unlike disposable subagents, NERO agents stay alive in memory with full conversation history, tool-use capabilities, and a shared memory bus.

Language: TypeScript (strict mode, ES2022 target, ESNext modules)
Runtime: Node.js >= 20.0.0
Transport: stdio (JSON-RPC 2.0) via @modelcontextprotocol/sdk
LLM Backend: Anthropic API via @anthropic-ai/sdk (Opus and Sonnet models)

Architecture Quick Reference

src/index.ts          Entry point, MCP server setup, tool registration
src/types.ts          All TypeScript interfaces and type definitions
src/agent.ts          NeroAgent class — tool-use loop, history management
src/agent-pool.ts     Agent lifecycle management (spawn/kill/reset/broadcast)
src/agent-tools.ts    7 internal tool definitions + executors (file I/O, grep, shell, memory)
src/memory-bus.ts     Shared key-value store with versioning
src/tools/*.ts        10 MCP tool handlers (one file per tool)

Build and Run

# Run from source (no build step needed)
npx tsx src/index.ts

# With file watching
npm run dev

# Type check (no emit)
npm run typecheck

# Full build to dist/
npm run build

Key Conventions

File naming: kebab-case.ts
Class naming: PascalCase (NeroAgent, MemoryBus, AgentPool)
MCP tool naming: nero_ prefix, snake_case (nero_spawn, nero_ask)
Internal tool naming: snake_case without prefix (read_file, run_command)
Tool registration pattern: Each MCP tool lives in src/tools/<name>.ts and exports a register*Tool() function
Zod schemas: Used for all MCP tool input validation
Logging: process.stderr.write() only (stdout is reserved for MCP JSON-RPC)

Important Constraints

stdout is exclusively for MCP JSON-RPC communication. Never use console.log().
Agent history must always start with a user role message (Anthropic API requirement).
The tool-use loop is capped at 15 iterations per nero_ask call.
History pruning uses both a sliding window (configurable, default 50 messages) and a token budget (100k estimated tokens).
Path sanitization blocks /proc, /sys, /dev, /boot for agent internal tools.
Destructive shell commands are blocked (rm -rf /, mkfs, dd if=, fork bombs).

Read the full file on GitHub · 79 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.

2d ago First seen · 79 lines · 822 tokens per session scan C 00706fc5e215

Subscribe to this mod's changes

agent-nero CLAUDE.md is an instructions file published in the GitHub repository sanchez314c/agent-nero (0 stars, last pushed 3mo ago), licensed MIT. It adds 822 tokens to every session, about $0.0041 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents LLM Memory and context ai-agents anthropic claude claude-code mcp

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

agent-nero CLAUDE.md