A working setup guide for headroom-ai: a local optimization proxy that compresses, caches, and streamlines every Claude API call before it leaves your machine. Zero changes to Claude Code itself.
This guide is headroom-specific. A similar setup exists with rtk-ai (different philosophy, overlapping outcomes). Both install cleanly via
uv tool install. Pick either — the wiring is the same shape.
| {# ========================= | |
| Merged Qwen Multimodal Chat Template from | |
| - https://github.com/allanchan339/vLLM-Qwen3-3.5-3.6-chat-template-fix | |
| - https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates | |
| Features: | |
| - developer role supported (from froggeric) | |
| - <|think_on|> / <|think_off|> toggles (from froggeric) | |
| - Historical reasoning HIDDEN by default (from allanchan339) | |
| - String-form tool arguments parsed as JSON (from allanchan339) | |
| - Non-ASCII escaped in tools/args JSON (from froggeric) |
Proxmox VE shell commands to create Debian 13 cloud-init ready VM template and deploy a VM with the template and cloud-init settings.
Ville Alatalo 2025 https://github.com/alatalo
| #!/usr/bin/env python3 | |
| """ | |
| Fix Claude Code session JSONL files affected by the whitespace-only text block bug. | |
| https://github.com/anthropics/claude-code/issues/23142 | |
| When a streaming response is interrupted before real content is emitted, Claude Code | |
| may freeze a whitespace-only text block (e.g. "\n\n" or "") into the conversation | |
| history. The API then rejects all subsequent requests with: | |
| 400 "messages: text content blocks must contain non-whitespace text" |
So you want to write a sync system for a web app with offline and realtime support? Good luck. You might find the following resources useful.
Database in a browser, a spec (Stepan Parunashvili)
What problem are we trying to solve with a sync system?
The web of tomorrow (Nikita Prokopov)
for 4.2.4 or higher, 4.2.5,4.2.6,4.3.7, it's works, this is the way which makes Always in evaluation mode.
Terminal, go to the dir : cd /Applications/Beyond Compare.app/Contents/MacOSBCompare to BCompare.bak: mv BCompare BCompare.bakBCompare , and chmod a+ux BCompare : touch BCompare && chmod a+ux BCompareBCompare with text editor, insert the script :#!/bin/bash
rm "/Users/$(whoami)/Library/Application Support/Beyond Compare/registry.dat"
"`dirname "$0"`"/BCompare.bak $@
| """ | |
| The most atomic way to train and run inference for a GPT in pure, dependency-free Python. | |
| This file is the complete algorithm. | |
| Everything else is just efficiency. | |
| @karpathy | |
| """ | |
| import os # os.path.exists | |
| import math # math.log, math.exp |
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.
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.
| import requests | |
| import re | |
| import sys | |
| from multiprocessing.dummy import Pool | |
| def robots(host): | |
| r = requests.get( | |
| 'https://web.archive.org/cdx/search/cdx\ | |
| ?url=%s/robots.txt&output=json&fl=timestamp,original&filter=statuscode:200&collapse=digest' % host) |