lguzzon

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

lguzzon

Self-Improving Claude Code: A Bootstrap Seed

The Hypothesis

A single prompt (~1400 tokens), placed in a project's .claude/CLAUDE.md, can bootstrap a Claude Code instance into a self-improving system — one that captures learnings, extracts patterns, evolves its own configuration, and gets meaningfully better at helping its user with each session.

No pre-built infrastructure required. No user-level config. No hooks, skills, templates, or elaborate folder hierarchies. Just a seed and the affordances Claude Code already provides.

Background

lguzzon

Using Claude Code with various AI providers

Create various settings file I have one file for each provider, all in ~/.claude

• KIMI K2.5: kimi_settings.json

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.moonshot.ai/anthropic",

lguzzon

#!/usr/bin/env bash
set -euo pipefail

bold() { printf "\033[1m%s\033[0m\n" "$*"; }
info() { printf "• %s\n" "$*"; }
warn() { printf "\033[33m! %s\033[0m\n" "$*"; }
err()  { printf "\033[31m✗ %s\033[0m\n" "$*" >&2; }
ok()   { printf "\033[32m✓ %s\033[0m\n" "$*"; }

require_macos() {

lguzzon

SPARC Framework Prompt Template

Introduction

You are an AI language model assisting in the development of a project using the SPARC framework, which consists of the following steps:

1. Specification
2. Pseudocode
3. Architecture
4. Refinement

lguzzon

# Create a new worktree and branch from within current git directory.
ga() {
  if [[ -z "$1" ]]; then
    echo "Usage: ga [branch name]"
    exit 1
  fi

  local branch="$1"
  local base="$(basename "$PWD")"
  local path="../${base}--${branch}"

lguzzon

Claude-SPARC Automated Development System For Claude Code

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

Overview

The SPARC Automated Development System (claude-sparc.sh) is a comprehensive, agentic workflow for automated software development using the SPARC methodology (Specification, Pseudocode, Architecture, Refinement, Completion). This system leverages Claude Code's built-in tools for parallel task orchestration, comprehensive research, and Test-Driven Development.

Features

lguzzon

{
  "mcpServers": {
    "count-r": {
      "command": "python",
      "args": [
        "C:\\Users\\mberm\\count-r-server\\server.py"
      ],
      "host": "127.0.0.1",
      "port": 8080,
      "timeout": 30000

lguzzon

from mcp.server.fastmcp import FastMCP
import time
import signal
import sys

# Handle SIGINT (Ctrl+C) gracefully
def signal_handler(sig, frame):
    print("Shutting down server gracefully...")
    sys.exit(0)

lguzzon

#!/usr/bin/env bash

set -euo pipefail

if [ "${1:-}" = "--help" ]; then
  cat << EOF
Usage: ijq [filename]

    A wrapper around jq that uses fzf to interactively build jq filters.