From f42668a37f8a9f396ecee5c277d52a92a45c94c0 Mon Sep 17 00:00:00 2001 From: Dan Wahlin Date: Wed, 1 Apr 2026 05:04:11 -0700 Subject: [PATCH 1/4] docs: add .claude/settings.json support and model picker reasoning effort MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Chapter 04: Add .claude/settings.json and .claude/settings.local.json to the Instruction File Formats table, plus a new 'Settings Files' section explaining how they differ from instruction files (added in CLI v1.0.12) - Chapter 01: Update the Switching Models section to describe the new full-screen model picker and inline reasoning effort adjustment with ← / → arrow keys (added in CLI v1.0.12) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --- 01-setup-and-first-steps/README.md | 4 +++- 04-agents-custom-instructions/README.md | 18 ++++++++++++++++++ 2 files changed, 21 insertions(+), 1 deletion(-) diff --git a/01-setup-and-first-steps/README.md b/01-setup-and-first-steps/README.md index 4b4e633..33f35db 100644 --- a/01-setup-and-first-steps/README.md +++ b/01-setup-and-first-steps/README.md @@ -464,11 +464,13 @@ Copilot CLI supports multiple AI models from OpenAI, Anthropic, Google, and othe copilot > /model -# Shows available models and lets you pick one. Select Sonnet 4.5. +# Opens a full-screen model picker. Select a model and press Enter. ``` > 💡 **Tip**: Some models cost more "premium requests" than others. Models marked **1x** (like Claude Sonnet 4.5) are a great default. They're capable and efficient. Higher-multiplier models use your premium request quota faster, so save those for when you really need them. +> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model. Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. + --- diff --git a/04-agents-custom-instructions/README.md b/04-agents-custom-instructions/README.md index a84d889..93355e0 100644 --- a/04-agents-custom-instructions/README.md +++ b/04-agents-custom-instructions/README.md @@ -423,9 +423,27 @@ Copilot will scan your project and create tailored instruction files. You can ed | `.github/copilot-instructions.md` | Project | GitHub Copilot specific | | `.github/instructions/*.instructions.md` | Project | Granular, topic-specific instructions | | `CLAUDE.md`, `GEMINI.md` | Project root | Supported for compatibility | +| `.claude/settings.json` | Project | Project-level behavior settings (committed, shared with team) | +| `.claude/settings.local.json` | Project | Personal settings override (not committed, stays on your machine) | > 🎯 **Just getting started?** Use `AGENTS.md` for project instructions. You can explore the other formats later as needed. +### Settings Files (.claude/settings.json) + +In addition to instruction files, Copilot reads **settings files** from your project's `.claude/` folder. These control *how* Copilot behaves rather than *what it knows* about your project. Think of instruction files as "here's our coding standards" and settings files as "here's how the tool should behave." + +- **`.claude/settings.json`** — Project settings committed to your repo, shared with the whole team. +- **`.claude/settings.local.json`** — Your personal overrides, not committed (add it to `.gitignore`). + +``` +your-project/ +└── .claude/ + ├── settings.json # committed — everyone gets these defaults + └── settings.local.json # NOT committed — personal overrides +``` + +> 💡 **Tip**: Both files are optional. Start with `AGENTS.md` for instructions, and only add a `settings.json` when you need to tune Copilot's behavior for the whole team. + ### AGENTS.md `AGENTS.md` is the recommended format. It's an [open standard](https://agents.md/) that works across Copilot and other AI coding tools. Place it in your repository root and Copilot reads it automatically. This project's own [AGENTS.md](../AGENTS.md) is a working example. From a9aa006dafd12ef7ea9fab765058329b55e8954f Mon Sep 17 00:00:00 2001 From: Dan Wahlin Date: Wed, 1 Apr 2026 09:07:54 -0700 Subject: [PATCH 2/4] Update reasoning effort explanation in README --- 01-setup-and-first-steps/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/01-setup-and-first-steps/README.md b/01-setup-and-first-steps/README.md index 33f35db..791ca68 100644 --- a/01-setup-and-first-steps/README.md +++ b/01-setup-and-first-steps/README.md @@ -469,7 +469,7 @@ copilot > 💡 **Tip**: Some models cost more "premium requests" than others. Models marked **1x** (like Claude Sonnet 4.5) are a great default. They're capable and efficient. Higher-multiplier models use your premium request quota faster, so save those for when you really need them. -> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model. Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. +> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model (if supported). Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. From bb34cc7a963eb2f81bdeb3b05f86b017d8adaa50 Mon Sep 17 00:00:00 2001 From: Dan Wahlin Date: Wed, 1 Apr 2026 09:09:39 -0700 Subject: [PATCH 3/4] Fix wording for reasoning effort explanation --- 01-setup-and-first-steps/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/01-setup-and-first-steps/README.md b/01-setup-and-first-steps/README.md index 791ca68..81fe4c4 100644 --- a/01-setup-and-first-steps/README.md +++ b/01-setup-and-first-steps/README.md @@ -469,7 +469,7 @@ copilot > 💡 **Tip**: Some models cost more "premium requests" than others. Models marked **1x** (like Claude Sonnet 4.5) are a great default. They're capable and efficient. Higher-multiplier models use your premium request quota faster, so save those for when you really need them. -> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model (if supported). Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. +> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model (when supported). Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. From 462fe7f5d26f72387e2822a113791270de2a95b2 Mon Sep 17 00:00:00 2001 From: Dan Wahlin Date: Wed, 1 Apr 2026 10:00:18 -0700 Subject: [PATCH 4/4] Update reasoning effort details in README.md --- 01-setup-and-first-steps/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/01-setup-and-first-steps/README.md b/01-setup-and-first-steps/README.md index 81fe4c4..145f942 100644 --- a/01-setup-and-first-steps/README.md +++ b/01-setup-and-first-steps/README.md @@ -469,7 +469,7 @@ copilot > 💡 **Tip**: Some models cost more "premium requests" than others. Models marked **1x** (like Claude Sonnet 4.5) are a great default. They're capable and efficient. Higher-multiplier models use your premium request quota faster, so save those for when you really need them. -> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model (when supported). Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.5 (high)`). Start with the default and raise it only when you need deeper analysis. +> 💡 **Reasoning effort**: Inside the model picker, use the ← and → arrow keys to adjust the **reasoning effort** for the selected model (when supported). Lower effort is faster; higher effort is more thorough. The current effort level is shown next to the model name in the session header (e.g., `claude-sonnet-4.6 (high)`). Start with the default and raise it only when you need deeper analysis.