Ben Vargas ben-vargas

Removing Codex Computer Use's built-in confirmation layers (macOS)

Instructions for a coding agent. Apply them on the user's Mac only when the user has asked for it. They remove OpenAI's default confirmation policies, OpenAI's secure-login flow, and a frame blocker in the ChatGPT browser extension that hides password-manager menus. After this, Codex follows the user's own instructions (AGENTS.md) instead of OpenAI's rules.

Tradeoff. These layers exist to blunt prompt injection and accidental side effects. Without them, the agent will send, delete, buy, and log in

Amp Orbs Running CLIProxyAPI + Amp Router Custom URL

This guide runs CLIProxyAPI in an Amp orb around the clock and publishes it at a stable public HTTPS URL through Tailscale Funnel. An Amp Custom URL model-routing connection then sends Claude requests there, and CLIProxyAPI serves them with your Claude Pro or Max subscription (OAuth login) instead of Amp credits.

The host scripts, templates, and an agent skill that walks you through every step live in ben-vargas/amp-plugins/skills/amp-cli-proxy-api. The easiest path is to install the skill and ask Amp to set it up:

amp skill add --global ben-vargas/amp-plugins/skills/amp-cli-proxy-api

Getting Started

Grok Build is a terminal-based AI coding assistant from xAI. It runs as a TUI (Terminal User Interface) that understands your codebase, executes shell commands, edits files, searches the web, and manages tasks -- all from your terminal.

You can use it interactively as a full-screen TUI, run it headlessly for scripting and CI/CD, or integrate it into editors and IDEs via the Agent Client Protocol (ACP).

Installation

Overriding Context Size for Opus 4.6 on GH Copilot OpenCode

Pi Harness Approach (current)

In ~/.pi/agent/models.json, models are defined with explicit contextWindow and maxTokens fields, plus a modelOverrides block for the github-copilot provider:

"modelOverrides": {
  "claude-opus-4.6": {
    "contextWindow": 192000,

@ccusage/omni Implementation Plan

Unified usage tracking across all AI coding assistants

Overview

Goal: Create a new @ccusage/omni package that aggregates usage data from all existing ccusage CLI tools into a single, unified view.

Initial Setup (New Projects)

This guide walks through setting up a new project with the bare git repository + worktree structure.

Setup Steps

1. Clone as Bare Repository

# Clone and setup bare repository

	{
	"diffMode": "github",
	"ideExtensionPromptedAt": {},
	"ideActivationNudgedForVersion": {},
	"enableCompletionBell": true,
	"completionSound": "fx-ok01",
	"awaitingInputSound": "fx-ack01",
	"soundFocusMode": "always",
	"commandAllowlist": [
	"ls",

	#!/usr/bin/env node
	// Minimal REST call to the Codex backend API using a local OAuth token.

	import fs from "fs";
	import os from "os";
	import path from "path";

	const authFile = process.env.CODEX_AUTH_FILE \|\| path.join(os.homedir(), ".codex", "auth.json");
	const auth = JSON.parse(fs.readFileSync(authFile, "utf8"));
	const token = auth.access_token \|\| auth.tokens?.access_token;

	#!/bin/bash
	# AI CLI tools wrapper for scheduled activation of usage windows
	# Outputs JSON for machine-readable logs with human-friendly summaries

	# ============================================================================
	# CONFIGURATION - Update these values for your environment
	# ============================================================================

	# System paths
	USER_HOME="/home/code"

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

	exec env RESTART_GATEWAY=1 /Users/clawd/clawd/scripts/patch-openclaw-codex-fast.sh "$@"