What should I use if I'm just starting with GitHub Copilot?

GitHub recommends starting with the GitHub Copilot CLI, which is a terminal interface with minimal UI to learn and offers direct, immediate interaction with the agent.

What is YOLO mode and why does it matter?

YOLO mode (also called "Allow All") lets the agent execute commands without asking permission for each action. It is critical because agents need autonomy to boost productivity; constant approval is a miserable user experience and trains you not to read what you are approving. However, you should only enable it when running agents in sandboxes (like GitHub Codespaces), not on your local machine, to protect private data.

Which AI model should I use for most work?

GitHub recommends using a medium-sized model such as GPT 5.6 Terra or Claude Sonnet with medium reasoning for most work, and sticking with the same model for the duration of a single feature or bug fix to preserve prompt caching and save tokens.

RoboticsGitHub Blog (AI)Jul 27, 2026Jul 28, 2026, 06:00 JST

The harness beats the tricks: GitHub's AI productivity recipe

Send on LINE

3 Key Points

What happened
GitHub shared a seven-step workflow for using GitHub Copilot effectively—starting with picking a tool, enabling autonomous execution (YOLO mode), prototyping with AI, planning methodically, implementing via Autopilot, iterating with human review, and finishing with a rubber-duck code review. The post argues that mastery of the core interface (the harness) and structured methodology matter far more than exotic prompts or obscure skills.
Why it matters
Many developers chase the latest AI tricks and plugins, but GitHub's evidence suggests the real productivity gain comes from understanding one tool deeply and using it systematically. For teams adopting Copilot, this means the bottleneck is workflow discipline and human judgment—knowing what quality looks like—not access to hidden techniques. It also points to a practical safety model: run agents with full autonomy in sandboxes (like GitHub Codespaces) rather than approving every action, since constant approval defeats the purpose and trains users to stop reading what they approve.
What to watch
The post specifically recommends using medium-sized models (GPT 5.6 Terra or Claude Sonnet) with medium reasoning, and sticking with one model per feature to maximize prompt caching savings. It also highlights the "grill-me" skill from Matt Pocock as a way to make the planning phase more thorough, and notes that GitHub Copilot's Autopilot phase orchestrates subagents automatically—using smaller models to explore files and larger models for complex tasks—without any manual configuration required.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The post emerges from a frustration with AI hype—the endless stream of "one weird trick" posts and tool announcements that promise mastery without substance. What GitHub is articulating is that productivity with AI agents comes not from novelty but from depth: learning one interface thoroughly, understanding its affordances, and applying a consistent methodology.

The seven-step workflow itself is grounded in familiar software practices: prototype before committing, plan edge cases, implement systematically, review rigorously, and iterate. What is new is that AI makes prototyping almost free—you can generate 20 design mockups in seconds—and planning can be crowdsourced to the model itself, which helps surface questions a human planner might miss. The author treats the model as a thought partner, not an oracle: the model asks questions, the human decides what matters, and the model implements. This division of labor—human judgment on quality and intent, AI on execution and ideation—appears to be where the real gain lies.

A second theme is the tension between autonomy and safety. YOLO mode (unrestricted command execution) is essential for productivity, but only safe in a sandbox. This suggests that the future of AI at work will depend less on perfect prompt engineering and more on infrastructure that permits trust: development containers, ephemeral environments, and clear boundaries between AI workspaces and production systems.

FAQ

What should I use if I'm just starting with GitHub Copilot?: GitHub recommends starting with the GitHub Copilot CLI, which is a terminal interface with minimal UI to learn and offers direct, immediate interaction with the agent.
What is YOLO mode and why does it matter?: YOLO mode (also called "Allow All") lets the agent execute commands without asking permission for each action. It is critical because agents need autonomy to boost productivity; constant approval is a miserable user experience and trains you not to read what you are approving. However, you should only enable it when running agents in sandboxes (like GitHub Codespaces), not on your local machine, to protect private data.
Which AI model should I use for most work?: GitHub recommends using a medium-sized model such as GPT 5.6 Terra or Claude Sonnet with medium reasoning for most work, and sticking with the same model for the duration of a single feature or bug fix to preserve prompt caching and save tokens.

GitHub Blog (AI)Read Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.