📰 Issue #23

Google Jules Launches CLI as Agent Wars Escalate

Google's Jules coding agent launched a CLI on October 2, embedding deeper into developer workflows as the battle for terminal dominance heats up.

Issue #23 - October 06, 2025 | 7-minute read

AI agents are leaving your IDE and moving into your terminal

👋 Hey there!

The AI agent wars just moved to your terminal. Google launched Jules CLI on October 2, GitHub shipped Copilot CLI updates with Claude 4.5 the next day, and both are racing to become infrastructure rather than tools. Meanwhile, even Jeff Bezos admits we're in an AI bubble—but says the tech is real. Here's what developers need to know this week.

💡 Google Jules Goes CLI—The Agent War Hits Your Terminal

On October 2, 2025, Google launched Jules Tools, bringing its AI coding agent directly into the command line (TechCrunch, October 2025). The CLI lets developers interact with Jules using terminal commands, eliminating context switching between web interfaces and GitHub. Just one day later on October 3, GitHub shipped Copilot CLI updates adding Claude Sonnet 4.5 support with enhanced model selection and image input capabilities (GitHub Changelog, October 2025). Google also made Jules' API public after previously reserving it for internal use only.

This isn't about convenience—it's about infrastructure. Jules CLI is designed for "scoped tasks" while Google's existing Gemini CLI handles "iterative collaboration," according to Kathy Korevec, director of product at Google Labs (TechCrunch, October 2025). The tools integrate with CI/CD pipelines, Slack, and terminal workflows, positioning agents as infrastructure rather than IDE accessories. GitHub is even deprecating its old gh-copilot extension by October 25, 2025, forcing users to migrate to the new agentic CLI. Jules offers a free tier with 15 daily tasks and 3 concurrent tasks, with paid plans at $19.99/month (Pro) and $124.99/month (Ultra) for higher limits.

The pattern is clear: coding agents are no longer content to live in your sidebar. They're embedding into build systems, terminals, and deployment pipelines. Google's senior developer advocate Denise Kwan explained that Jules executes tasks independently once users approve its plan—less interactive by design (Medium, October 2025). Meanwhile, GitHub's Copilot CLI now supports switching between AI models mid-session using the /model command, with Claude Sonnet 4.5 gradually rolling out to Pro, Pro+, Business, and Enterprise users. The CLI wars signal a shift from assistive tooling to foundational infrastructure.

Bottom line: AI agents are moving from helpful sidekicks to core infrastructure—and your terminal is the new battleground.

🛠️ Tool Updates

GitHub Copilot CLI with Claude Sonnet 4.5 - Terminal-native coding agent with multi-model support

GitHub rolled out major Copilot CLI enhancements on October 3, 2025, adding Claude Sonnet 4.5 support with the ability to switch models mid-session using /model commands (GitHub Changelog, October 2025). The update includes image input support, enhanced tool permissions using glob patterns, and improved usage tracking with the /usage command. Available for Pro, Pro+, Business, and Enterprise users, with the old gh-copilot extension being deprecated by October 25, 2025. Pricing remains tied to existing Copilot subscription tiers.

Microsoft Agent Framework - Unified SDK merging AutoGen and Semantic Kernel

Microsoft launched Agent Framework in public preview on October 1-2, 2025, unifying AutoGen's innovative orchestration with Semantic Kernel's enterprise stability (Microsoft Azure Blog, October 2025). The open-source framework for .NET and Python enables building multi-agent systems in under 20 lines of code, with support for Model Context Protocol (MCP) and Agent2Agent (A2A) communication. Early enterprise adopters include KPMG, Commerzbank, and Citrix.

Claude for Chrome Extension - Browser-native Claude with computer use capabilities

Anthropic began rolling out Claude for Chrome to Max plan waitlist users in late September 2025, enabling browser automation and computer use directly in Chrome (Anthropic, September 2025). The extension leverages Claude Sonnet 4.5's enhanced computer use abilities.

💰 Cost Watch

CLI Pricing Wars: Google Jules CLI launched with aggressive free tier pricing—15 daily tasks free, then $19.99/month for Pro (5x usage) and $124.99/month for Ultra (20x usage) as announced October 2, 2025 (TechCrunch, October 2025). GitHub Copilot CLI remains bundled with existing subscriptions at $10/month individual or $19/month business.

💡 Money-saving insight: If you're testing AI coding agents, start with Jules' free tier (15 daily tasks) before committing to paid plans. For teams already on GitHub Copilot, the CLI is included—no additional cost.

🔧 Quick Wins

🔧 Try Jules CLI for Terminal Workflows: Install Google Jules CLI and delegate scoped terminal tasks like bug fixes or feature implementations without leaving your command line. The free tier gives you 15 daily tasks and 3 concurrent operations—enough to test if terminal-native agents fit your workflow. Takes 10 minutes to set up and can save hours on context switching.

🎯 Switch GitHub Copilot to Claude Sonnet 4.5: Update GitHub Copilot CLI to the latest version (v0.0.329+) and use the /model command to test Claude Sonnet 4.5 for complex coding tasks. The model excels at longer-horizon work and maintaining context across multi-step implementations. Available now for Pro, Pro+, Business, and Enterprise users.

⚡ Migrate from gh-copilot Before October 25: GitHub is deprecating the old gh-copilot CLI extension on October 25, 2025. Update to the new GitHub Copilot CLI now to avoid workflow disruptions. Run npm install -g @github/copilot@latest to migrate—it takes 2 minutes and prevents future breakage.

🌟 What's Trending

💸 Bezos Calls It—"We're in an Industrial AI Bubble"

Jeff Bezos declared AI is in an "industrial bubble" on October 3, 2025. Goldman Sachs CEO agreed: "There will be a reset." AI investment hit $252.3 billion in 2024, projected $400 billion for 2025. MIT research shows 95% of organizations get zero return from GenAI.

The warnings aren't from skeptics—they're from the billionaires building it. When Bezos says "every idea gets funded, good and bad," he's describing his own fundraising climate. Read more →

🔧 Microsoft Ends the Framework Wars

Microsoft launched Agent Framework on October 1-2, 2025, merging AutoGen and Semantic Kernel into one SDK. Developers were stuck choosing between AutoGen (experimental) and Semantic Kernel (limited). The new framework combines both with MCP, Agent2Agent, and OpenAPI support. Build agents in under 20 lines of code.

This is Microsoft admitting the multi-framework strategy failed. Instead of forcing devs to pick sides, they hit reset. The agent infrastructure is consolidating. Read more →

💬 Are CLI agents changing your workflow?

Have you tried Jules CLI, GitHub Copilot CLI, or other terminal-native coding agents? Are they actually useful, or just more noise in your terminal? Hit reply—I read every message and I'm curious about your real-world experience.

— Pranay, INFOLIA AI

Missed Issue #22? Catch up here →