Welcome back to Agentic Coding Weekly. Here are the updates on agentic coding tools, models, and workflows for the week of Mar 8-14, 2026.
Executive Summary:
Opus 4.6 and Sonnet 4.6 now get the full 1M context window in Claude Code at standard pricing.
Anthropic is temporarily doubling five-hour usage for Claude and Claude Code during off-peak hours on most non-Enterprise plans.
New
/btwcommand in Claude Code, voice mode rolled out to all users, and new agent team-based code review.Gemini CLI now has Plan mode.
JetBrains Air, a new agentic development environment, is now in public preview on macOS.
Worth reading this week: a practical site for checking which local models your machine can run, 8 levels of agentic engineering, and Simon Willison on using AI to improve code quality instead of just shipping faster.
Meanwhile, this screenshot got a lot of attention on Hacker News where Opus 4.6 in OpenCode asks the user "Shall I implement it?". The user says "no". Opus proceeds to implementation anyway.
1. Tooling and Model Updates
1M Context for Opus 4.6 and Sonnet 4.6
Claude Opus 4.6 and Sonnet 4.6 now include the full 1M context window at standard pricing. No long-context premium. Claude Code Max, Team, and Enterprise users on Opus 4.6 will default to 1M context automatically. Check the announcement.
Voice Mode in Claude Code
Voice mode has rolled out to all users. /voice to toggle, hold space to talk, release to send. Works like push-to-talk and the transcript gets inserted at the cursor position.
I compared it with what I typically use, gpt-4o-transcription model through VoiceInk. Claude code voice mode seems to struggle with technical terms and overall quality of transcription is lower.
Claude March 2026 Usage Promotion
From March 13 through March 28, five-hour usage for Claude (including Claude Code) is doubled during off-peak hours, that is, any time outside 12-6 PM GMT on weekdays. Applies to Free, Pro, Max, and Team plans. Enterprise not included. Check the announcement.
Quick Updates
Claude Code now has an agent team-based code review system. Multiple agents coordinate to review a PR from different angles. Anthropic says it generally averages $15-25 per PR.
New /btw command in Claude Code allows to ask a quick question about the current work without adding to the conversation history and thus keeping the main context clean.
Plan mode is now available in Gemini CLI.
JetBrains launched Air, an agentic development environment (ADE), as a public preview for macOS. This is JetBrains building a purpose-built environment for agentic workflows rather than adding agent features onto an existing IDE.
2. Community Picks
The 8 Levels of Agentic Engineering
Useful mental model: Tab Complete -> Agent IDE -> Context Engineering -> Compounding Engineering -> MCP and Skills -> Harness Engineering -> Background Agents -> Autonomous Agent Teams. Read the post and discussion on HN.
MCP is Dead; Long Live MCP!
An argument for why local MCP over stdio doesn't offer much over plain CLI tools, but remote MCP over HTTP is still the right tool for organizations. Read the post and discussion on HN.
AI Should Help Us Produce Better Code
New chapter in the Simon Willison's Agentic Engineering Patterns guide. Read the post.
If we want the best results from our agents, we should aim to continually increase the quality of our codebase over time. Small improvements compound. Quality enhancements that used to be time-consuming have now dropped in cost to the point that there's no excuse not to invest in quality at the same time as shipping new features. Coding agents mean we can finally have both.
Can I Run AI Locally?
A website that automatically tells you what open-weight models you can run locally based on your machine specs, along with expected tokens per second. Check the website.
That’s it for this week. I write this weekly on Mondays. If this was useful, subscribe below:
