diff --git a/website/src/content/docs/learning-hub/agents-and-subagents.md b/website/src/content/docs/learning-hub/agents-and-subagents.md index 6d85ee8cb..3272a7334 100644 --- a/website/src/content/docs/learning-hub/agents-and-subagents.md +++ b/website/src/content/docs/learning-hub/agents-and-subagents.md @@ -3,7 +3,7 @@ title: 'Agents and Subagents' description: 'Learn how delegated subagents differ from primary agents, when to use them, and how to launch them in VS Code and Copilot CLI.' authors: - GitHub Copilot Learning Hub Team -lastUpdated: 2026-07-01 +lastUpdated: 2026-08-05 estimatedReadingTime: '9 minutes' tags: - agents @@ -134,6 +134,20 @@ The important behavior is different from a single chat turn: That makes `/fleet` a practical way to launch subagents even if you are not authoring custom agent files yourself. +### Sessions sidebar for parallel work *(v1.0.76+, experimental)* + +When you're running multiple independent tasks in Copilot CLI, the **Sessions sidebar** gives you a split-view TUI panel to manage concurrent sessions without leaving the terminal. You can switch between active sessions, spawn new ones, and see their status at a glance — all without starting a new terminal window. + +Enable it with experimental mode: + +``` +/experimental on +``` + +Once enabled, the Sessions sidebar appears as a panel alongside your current conversation. Each session tracks its own conversation history, working directory, and model selection independently. This complements `/fleet` — while `/fleet` orchestrates subagents within a single session, the Sessions sidebar lets you manage fully independent parallel workstreams (for example, one session refactoring a module while another writes tests for a different service). + +> **Tip**: In v1.0.78+, switching sessions no longer restarts MCP servers or rebuilds your hook state — the switch is nearly instant and the session resumes exactly where it left off. + ### Rubber-duck agent Available in `/experimental` (v1.0.42+), the **rubber-duck agent** applies a novel multi-model pattern: when you're working in a GPT-powered session, the rubber-duck agent internally routes certain requests through Claude to provide a second perspective. The idea is similar to rubber-duck debugging — talking through a problem with a different "listener" often surfaces assumptions or blind spots you didn't notice. @@ -206,7 +220,7 @@ Yes, when the delegated worker is a custom agent with its own frontmatter. **Are subagents always parallel?** -No. They can run sequentially when one step depends on another, or in parallel when work items are independent. +No. They can run sequentially when one step depends on another, or in parallel when work items are independent. In v1.0.76+, the CLI includes improved subagent delegation that makes it better at automatically choosing parallelism for small, independent tasks — you no longer need to explicitly request parallel execution in many common scenarios. **Can I control how many subagents run simultaneously?** diff --git a/website/src/content/docs/learning-hub/github-copilot-app.md b/website/src/content/docs/learning-hub/github-copilot-app.md index b32fe072f..6870a2457 100644 --- a/website/src/content/docs/learning-hub/github-copilot-app.md +++ b/website/src/content/docs/learning-hub/github-copilot-app.md @@ -3,7 +3,7 @@ title: 'Getting Started with the GitHub Copilot app' description: 'Learn about the GitHub Copilot app, a desktop experience built for agent-native development. Understand its key features and who it''s for.' authors: - GitHub Copilot Learning Hub Team -lastUpdated: 2026-06-17 +lastUpdated: 2026-08-05 estimatedReadingTime: '8 minutes' tags: - copilot-app @@ -67,6 +67,17 @@ Each session the Copilot app creates runs in its own **git worktree**—a real, This makes it easy to dispatch multiple agents and trust they won't interfere with each other. +#### Creating a new worktree from the CLI *(v1.0.78+, experimental)* + +If you prefer terminal-first workflows, the Copilot CLI exposes `/new-worktree` as an experimental command that mirrors the Copilot app's worktree creation — it creates a new git worktree and opens a fresh conversation in it, without leaving the CLI: + +``` +/experimental on +/new-worktree +``` + +This is useful when you want the isolation of separate worktrees for parallel work but are working in the terminal rather than the desktop app. + ### Canvases **Canvases** are interactive work surfaces where you and agents collaborate. Instead of long chat threads, a canvas shows the actual work: diff --git a/website/src/content/docs/learning-hub/understanding-mcp-servers.md b/website/src/content/docs/learning-hub/understanding-mcp-servers.md index 3e02a1426..72d874711 100644 --- a/website/src/content/docs/learning-hub/understanding-mcp-servers.md +++ b/website/src/content/docs/learning-hub/understanding-mcp-servers.md @@ -3,7 +3,7 @@ title: 'Understanding MCP Servers' description: 'Learn how Model Context Protocol servers extend GitHub Copilot with access to external tools, databases, and APIs.' authors: - GitHub Copilot Learning Hub Team -lastUpdated: 2026-07-13 +lastUpdated: 2026-08-05 estimatedReadingTime: '8 minutes' tags: - mcp @@ -317,6 +317,8 @@ You can also open the `/mcp` manager while the agent is working to toggle server **Toggling servers on and off** (v1.0.66+): From the `/mcp` list view, you can **enable or disable individual MCP servers** without editing your config file. Select a server in the list and toggle it — disabled servers won't start in future sessions and their tools won't be available to agents. This is useful for temporarily disabling a server that's causing slowdowns or errors without removing it from your configuration entirely. +**Session switching preserves MCP connections** *(v1.0.78+)*: When you switch between sessions in the Sessions sidebar, MCP servers are no longer restarted. The server processes remain running and attached — the switch is instant, and each session resumes with its existing server connections. Previously, switching sessions would restart all local MCP servers and rebuild hook state, which could interrupt long-running server processes. + **Common causes and fixes**: | Symptom | Likely Cause | Fix | diff --git a/website/src/content/docs/learning-hub/using-copilot-coding-agent.md b/website/src/content/docs/learning-hub/using-copilot-coding-agent.md index 4ff5a3441..cf85123aa 100644 --- a/website/src/content/docs/learning-hub/using-copilot-coding-agent.md +++ b/website/src/content/docs/learning-hub/using-copilot-coding-agent.md @@ -3,7 +3,7 @@ title: 'Using the Copilot Coding Agent' description: 'Learn how to use GitHub Copilot coding agent to autonomously work on issues, generate pull requests, and automate development tasks.' authors: - GitHub Copilot Learning Hub Team -lastUpdated: 2026-05-13 +lastUpdated: 2026-08-05 estimatedReadingTime: '12 minutes' tags: - coding-agent @@ -377,6 +377,16 @@ Since v1.0.47, `--resume` also surfaces **cloud agent sessions that haven't yet > **Note**: Remote control replaces the earlier "steering" feature. If you see references to steering in older documentation, remote control is the updated equivalent. +### Switching approval modes with `/permissions` *(v1.0.78+)* + +When you're connected to a running session — either locally or via remote control — you can switch between **interactive**, **plan**, and **autopilot** approval modes on the fly without restarting the session: + +``` +/permissions # open the approval mode picker +``` + +This lets you tighten or relax what the agent can do autonomously mid-session. For example, you might start in interactive mode to review early steps, then switch to autopilot once you're confident the agent is on the right track. Switching to plan mode pauses new file writes and lets you review the agent's intended changes before they land. + ## Hooks and the Coding Agent Hooks are especially valuable with the coding agent because they provide deterministic guardrails for autonomous work: