What is HARTOS CLAUDE.md?

Repository instructions for HART OS, a distributed platform that coordinates autonomous AI agents and crowdsourced computing. Its recipe system can learn a task once and replay it later with fewer model calls.

Use them when modifying HART OS, following its master-key and exclusion-zone rules, or locating canonical sibling repositories and project files.

They describe the project's purpose, safety controls, required branch practices, and relationships with related repositories.

How hertz-ai/HARTOS configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to hertz-ai/HARTOS. 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/hertz-ai/HARTOS/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/hertz-ai/HARTOS

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/hertz-ai/hartos/claude-md/github.svg)](https://agentmods.dev/instructions/hertz-ai/hartos/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/hertz-ai/hartos/claude-md"><img src="https://agentmods.dev/badge/instructions/hertz-ai/hartos/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/hertz-ai/hartos/claude-md"><img src="https://agentmods.dev/badge/instructions/hertz-ai/hartos/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 9,346 This file is loaded in full into every session.

When invoked 9,346 The same file — it is already loaded in full.

Security scan A 2 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.

Model	Per session	Once invoked
Fable 5.1	$0.09346	$0.09346
Opus 5.5	$0.03738	$0.03738
Sonnet 5.5	$0.01869	$0.01869
Haiku 4.5	$0.00935	$0.00935

Measured 20d ago against content hash 98cafe0c3ac3, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

HARTOS CLAUDE.md scanned grade A 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 20d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- `requests.get/post` / `urllib.urlopen` / `socket.connect` → MUST

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

- `os.popen` / `os.system` without timeout → BANNED (Rule 5 of

CLAUDE.md · 666 lines

How it starts

The opening of the file, as written. The whole thing — 666 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.

RULE 0 — NEVER ASSUME (ranked above every other rule in this file)

Owner ruling 2026-09-05. Outranks every protocol below and any ship-mode urgency. It is not advice.

Never assume. If a measurement of the exact claim does not exist, the claim is not made. "Probably", "consistent with", "the logs suggest" are assumptions wearing a hedge. When unmeasured, the answer is "I have not checked".
Grep only LOCATES. A grep hit is a pointer, never evidence, and never a basis for a conclusion or an edit.
Read the FULL file — every file the grep matched, end to end. Not a window around the hit. Partial reads are the documented cause of this codebase's wrong diagnoses and parallel-path regressions.
Verify to the user-visible outcome, not the furthest log line reached. A pipeline that "ran" has produced nothing until the user's actual result is read back.

Measured cost of breaking this (single session, 2026-09-05): six conclusions asserted then withdrawn — a real registered tool called "fictional"; a missing log marker read as "no tool executed" when those tools cannot emit it; and a 179-second reuse turn analysed entirely from internal logs while the response the user actually received said "Your local AI is busy with another task right now."

Full rule + evidence table: Nunba memory/feedback_never_assume.md.

Project Overview

HART OS - Hevolve Hive Agentic Runtime

Crowdsourced compute infrastructure that orchestrates fully autonomous Hive AI Training. Open multi-agent platform democratizing access to the best intelligence, just like the internet is open, this will be open. Key innovation: Recipe Pattern - learn task execution once (CREATE mode), then replay efficiently (REUSE mode) without repeated LLM calls.

What it does: Enables distributed thought processes across crowdsourced agents, distributed coding agents that incentivize compute providers, and a self-sustaining economic model (like a vending machine, but at national economy scale) so that no single entity, no government, and no private corporation ever monopolizes AI. Intelligence belongs in the hands of the common person.

Read the full file on GitHub · 666 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.

20d ago Changed · +44 lines · +862 tokens per session 98cafe0c3ac3
2mo ago Changed · +27 lines · +350 tokens per session 8f24dc5d5c83
2mo ago First seen · 595 lines · 8,134 tokens per session scan A ee25d8dd031e

Subscribe to this mod's changes

HARTOS CLAUDE.md is an instructions file published in the GitHub repository hertz-ai/HARTOS (51 stars, last pushed 2d ago), licensed Apache-2.0. It adds 9,346 tokens to every session, about $0.0374 per session on Opus 5.5. A static security scan graded it A with 2 findings (makes network calls, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

agent-framework agentic-ai ai-agent ai-runtime chatbot federated-learning

HARTOS CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
HARTOS CLAUDE.mdhertz-ai/HARTOS	Instructions file	51	2d ago	Apache-2.0	A	9,346	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	yesterday	MIT	A	7,461	Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	yesterday	MIT	A	4,680	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	5,516	GitHub Copilot
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	yesterday	MIT	A	3,093	GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness	Instructions file	274k	2d ago	MIT	A	4,313	Codex, OpenCode

HARTOS: Instructions file for Claude Code