CORE CLAUDE.md

CORE CLAUDE.md is an instructions file for coding agents from DariuszNewecki/CORE. It costs 8,080 tokens per session, scanned A, original, MIT.

A set of project instructions for Claude Code, an AI programming assistant, explaining CORE’s structure, rules, and working process.

In plain words
What is it for?
Use it when working on CORE, a software factory that supervises AI-generated code under human-defined rules and verification.
Why use it?
It tells the assistant which files define requirements, architecture, and implementation boundaries before it changes code.

Instructions file

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/dariusznewecki/core/claude-md
Clone the repo
git clone --depth 1 https://github.com/DariuszNewecki/CORE

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 CORE CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/dariusznewecki/core/claude-md.svg)](https://agentmods.dev/instructions/dariusznewecki/core/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/dariusznewecki/core/claude-md"><img src="https://agentmods.dev/badge/instructions/dariusznewecki/core/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 8,080 This file is loaded in full into every session.
When invoked 8,080 The same file — it is already loaded in full.
Security scan A 0 findings. 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.

ModelPer sessionOnce invoked
Fable 5 $0.08080 $0.08080
Opus 5 $0.04040 $0.04040
Sonnet 5 $0.01616 $0.01616
Haiku 4.5 $0.00808 $0.00808

Measured 3d ago against content hash ccd547f0865a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

CORE 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 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.

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 · 512 lines

How it starts

The opening of the file, as written. The whole thing — 512 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — CORE

Loaded automatically at the start of every Claude Code session. Read fully before touching code.


What CORE is, and what you are

CORE is a constitutionally-governed software factory: a runtime that supervises AI code generation under deterministic rules. .intent/ is law (read by the runtime); .specs/ is architectural reasoning (read by humans); src/ is the implementation.

You (Claude Code) are the execution arm, not the governor. The human owns this repo, writes intent, and reviews outputs — they do not write code manually. Your job is to produce correct, complete files that conform to .intent/ and the decisions in .specs/decisions/. AI output is not trusted by default — it is verified. Produce work that earns the verification.

This file is a development contract, not a runtime governance posture. CORE is bootstrapped on itself; the restrictions here govern how Claude Code works on this repo and are intentionally permissive under governor confirmation. The strong version — what CORE enforces on governed projects at runtime — lives in .intent/. Do not import that live strictness into this dev contract, and do not export this dev permissiveness into runtime rules.


Source layout

src/
  api/        — FastAPI routes and dependency providers only. No business logic.
  body/       — Analyzers, atomic actions, services, infrastructure workers. Execution layer.
  cli/        — Typer CLI commands. Rich rendering allowed here only.
  mind/       — Constitutional logic engines (ast_gate, glob_gate, llm_gate, runtime_gate, …).
                Reads .intent/ at runtime. No I/O, no execution, no Body/Will invocation.
  shared/     — Cross-cutting substrate. Must not import src/mind/, src/body/, or src/will/.
  will/       — Autonomous developer, cognitive orchestration, agents.
.intent/      — Governance law as data (YAML/JSON). Read at runtime via IntentRepository;
                never imported as Python.
.specs/       — Charter, northstar, papers/ (47+), requirements/ (URS), decisions/ (ADRs),
                planning/ (roadmaps).

Read the full file on GitHub · 512 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. 3d ago First seen · 512 lines · 8,080 tokens per session scan A ccd547f0865a

Subscribe to this mod's changes

CORE CLAUDE.md is an instructions file published in the GitHub repository DariuszNewecki/CORE (38 stars, last pushed 4d ago), licensed MIT. It adds 8,080 tokens to every session, about $0.0404 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 2026-08-30.