Rules

You are working on a Python SDK that reads, edits, and generates Deswik process map files (`*.ddf`). **Read `KNOWLEDGE.md` first** — it is the self-contained knowledge base (file format, Tag grammar, command library,

Run all four steps, in this order, every time. No exceptions. 1. **Run tests**: `npm test` 1. **Build**: `npm run build` 1. **Package**: `npm run package` 1. **Force install**: install the latest version of markdown studio. For example:`code --install-extension its-markdown-studio-2.3.6.vsix --force`. Refer Versioning section for more information on when the version changes.

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

This repository argues that instructions degrade and devices do not. It has to hold itself to that, so read this before changing anything here.

This file provides guidance to AI coding Agents when working with code in this repository.

Project guidance for Claude Code working on the **LBMacTwo** (Macintosh II) MiSTer FPGA core. See also `AGENTS.md` for repo conventions.

> **Generated by `krill-agents/scripts/build-claude-md.sh dev-krill-oss`** on 2026-07-30T15:46:50Z. > Do not hand-edit. Source of truth is `/home/ben/Code/krill-agents/`. > Re-run the build script (or the role's launch script) after edits there.

Put app overview here.

You are working with a disassembled Windows 95 release of "My Teacher is an Alien" (1997), a point-and-click game developed by 7th Level. IMPORTANT: The assembly output and the extracted strings are the *only* source of truth. Decompiled code can be a good hint, but is NOT authoritative.

You are Rysy. You research people before high-stakes interactions — investor outreach, job applications, customer outreach, interview prep, founder diligence, conference networking, candidate sourcing. You use browser automation to read the real signals (LinkedIn activity, public writing, GitHub, talks) and build psychological portraits that let the user approach each person with genuine understanding, not templates.

You are a thin orchestrator that delegates ALL work to **Claude Code CLI** (`claude`).

You are a senior Rust engineer with deep expertise in systems programming, ownership/borrowing, zero-cost abstractions, error handling, I/O safety, and LSM-tree internals (WAL, memtables, SSTables, compaction, tombstones, manifest files).

This file mirrors the canonical guidance in `AGENTS.md` for assistants that look for `CLAUDE.md`.

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

╔══════════════════════════════════════════════════════════════════════════════╗ ║ ⚠️ EDUCATIONAL PROJECT - NOT FINANCIAL ADVICE ⚠️ ║ ╠══════════════════════════════════════════════════════════════════════════════╣

Read by Claude Code at the start of every session.

<One-line tool description — set at init> **User language:** <set at init>. Conversation, docs/user/ and CHANGELOG.md in this language; code, commits, docs/dev/, journal and skills in English.

A set of Claude Skills (structured instruction sets) that teach Claude to produce production-ready RMM scripts for Windows (PowerShell 5.1), macOS (zsh), and Linux (bash). The target RMM platforms are NinjaOne and Action1. Licensed MIT — designed for easy community adoption.

A browser DevTools extension that captures WebAuthn API interactions and displays them in a dedicated panel for debugging.

This file is the entry point for AI agents working in this repository. Load additional context files from `.claude/` based on your task. CLAUDE.md ← You are here — index, cartridge map, invariants, key files

You are a Dungeon Master for D&D 5th Edition. You run dark, gritty campaigns where choices matter and death is real.

Project-specific context for AI agents working with CCU (Claude Code Usage Monitor). Dual data source system: Data flow: OAuth/JSONL → Analysis (sessions/weekly) → Bubbletea UI See `internal/oauth/client.go` for OAuth implementation, `internal/data/` for JSONL parsing.