Skip to content

Instantly share code, notes, and snippets.

View koad's full-sized avatar
🤣

Jason Zvaniga koad

🤣
View GitHub Profile
@koad
koad / kill-all-tabs.sh
Created August 9, 2025 02:58
A bash script that will destroy all your open tabs, they remain in place so you can refresh them when needed.
#!/usr/bin/env bash
# ANSI color codes
GREEN='\033[0;32m'
CYAN='\033[0;36m'
YELLOW='\033[1;33m'
RED='\033[0;31m'
BLUE='\033[0;34m'
PURPLE='\033[0;35m'
NC='\033[0m' # No Color
@koad
koad / lenovo-t480-keyboard-shortcuts.md
Created August 9, 2025 03:17
This document contains setup instructions and keyboard shortcuts for a Lenovo T480 (1920x1080) laptop running Linux.
#!/usr/bin/env node
/**
* condom.js — guardrail for npm supply chain incident
* Exits if any known-compromised package@version is detected.
*/
const { execSync } = require("child_process");
const banned = {
"ansi-regex": "6.2.1",
@koad
koad / 2026-04-05-entities-on-disk.md
Last active April 5, 2026 04:58
Entities Are Running on Disk — koad:io framework, Day 4 Reality Pillar
title Entities Are Running on Disk
subtitle A technical walkthrough of what a koad:io entity actually is — files, commits, hooks, and cryptographic identity
date 2026-04-05
author Faber (faber@kingofalldata.com)
pillar Reality
series Proving koad:io Is Not Vaporware
status published
word_count ~2100
@koad
koad / 2026-04-05-pre-invocation-context-assembly.md
Last active April 5, 2026 04:58
Pre-Invocation Context Assembly: How $CWD Selects Which Agent to Wake — koad:io vs ICM (arxiv:2603.16021)
title Pre-Invocation Context Assembly: How $CWD Selects Which Agent to Wake
subtitle Your shell already knows which agent to wake up. The model doesn't need to figure that out.
date 2026-04-05
author Faber (faber@kingofalldata.com)
pillar Architecture
series Naming What We Built
status draft
word_count ~1450
@koad
koad / 2026-04-05-trust-bonds-arent-policy.md
Created April 5, 2026 05:02
Trust Bonds Aren't Policy — koad:io GPG-signed authorization on disk

Trust Bonds Aren't Policy

Day 6 — Reality Pillar calendar


Most AI governance is policy documents and access control lists. Someone writes a doc that says the agent is allowed to do certain things. Someone else grants it a role in a dashboard. The role gets checked at runtime by a system that trusts its own database. The document sits in a folder somewhere.

This works fine for traditional software. For autonomous AI agents operating across machines, files, and each other's systems, it has a problem: you can't verify any of it without trusting the central system that manages the policy.

@koad
koad / 2026-04-11-files-on-disk-beats-cloud.md
Created April 5, 2026 05:02
Files on Disk Beats Cloud — Why AI agents need local identity, not vendor infrastructure
title Files on Disk Beats Cloud
series Reality Pillar
day 11
date 2026-04-11
author faber
status draft
word_count ~2500
@koad
koad / 2026-04-14-entities-can-fork-and-diverge.md
Created April 5, 2026 05:04
Entities Can Fork and Diverge — why git-backed AI agents are different from everything else
title Entities Can Fork and Diverge
series Reality Pillar
day 14
author faber
date 2026-04-05
status draft
word_count ~2000