ntk CLAUDE.md

Project instructions for NTK covering its implementation sequence, project-specific rules, skills, testing, and contribution process.

In plain words
What is it for?
Use them when developing NTK, especially for implementation tasks, Rust changes, testing, security checks, or contributor documentation.
Why use it?
They give coding agents the repository’s required order of work and project conventions before making changes.

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/luizinhoh2o1/ntk/claude-md
Clone the repo
git clone --depth 1 https://github.com/luizinhoh2o1/ntk
Per session 6,904 This file is loaded in full into every session.
When invoked 6,904 The same file — it is already loaded in full.
Security scan B 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.

ModelPer sessionOnce invoked
Fable 5 $0.06904 $0.06904
Opus 5 $0.03452 $0.03452
Sonnet 5 $0.01381 $0.01381
Haiku 4.5 $0.00690 $0.00690

Measured yesterday against content hash d6363362454c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

ntk CLAUDE.md scanned grade B 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 yesterday.

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.

2. Patch idempotente em `~/.claude/settings.json` adicionando o hook `PostToolUse`
CLAUDE.md · 650 lines

How it starts

The opening of the file, as written. The whole thing — 650 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.

Project status & community

NTK is an open initiative. The current maintainer can only cover so much ground — we actively need collaborators for: stack-trace classifier coverage (more languages/frameworks), real-hardware benchmarks on GPUs we don't own (AMD / Apple Silicon / Intel AMX), L3 prompt A/B testing, hook compatibility with editors other than Claude Code & OpenCode, and translations of the docs site. Issues and PRs are welcome at https://github.com/VALRAW-ALL/ntk. See CONTRIBUTING.md for the concrete starter tasks.

Implementation Guide

Implementando do zero? Siga IMPLEMENTATION-SEQUENCE.md — 26 etapas ordenadas por dependência, com critério de "pronto" em cada uma. Não pule etapas.

Project-specific rules (.claude/rules/)

These rules are loaded automatically when you work on matching files:

Rule Triggers on
implementation-gate.md every code change — security, memory, quality, issue fidelity checklist
clippy-gate.md any Rust change — the CI clippy flags that must pass locally
cuda-ci.md edits to .github/workflows/release.yml with CUDA steps
gpu-vendor.md src/gpu.rs, src/config.rs, src/main.rs setup wizard, layer3_llamacpp.rs
l1-l2-invariants.md any L1/L2 algorithm change — the 5 invariants CI enforces
l1-template-dedup.md group_by_template changes — blank-line handling
stack-trace-classifier.md is_framework_frame extensions, new language support
l4-context-injection.md layer4_context.rs, server handler, hook scripts

Project-specific skills (.claude/skills/)

Skill When to invoke
bench-runner Running microbench / full bench / prompt A/B / Claude Code A/B
add-stack-trace-language Extending L1's classifier to a new language/framework

Skills e Agentes Disponíveis neste Projeto

Read the full file on GitHub · 650 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. yesterday First seen · 650 lines · 6,904 tokens per session scan B d6363362454c

Subscribe to this mod's changes

ntk CLAUDE.md is an instructions file published in the GitHub repository luizinhoh2o1/ntk (10 stars, last pushed 4mo ago), licensed MIT. It adds 6,904 tokens to every session, about $0.0345 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.