docs: update Learning Hub for Copilot CLI v1.0.72-v1.0.75 changes

- Add Claude Opus 5 (v1.0.75) and gemini-3.6-flash (v1.0.74) to the model selection guide in building-custom-agents.md - Document agentStop hook stop_hook_active flag and 8-consecutive-block loop-prevention behavior (v1.0.72) in automating-with-hooks.md - Add Open Plugin Spec v1 and mcp.json support note (v1.0.74) in installing-and-using-plugins.md - Document /model plan command for per-mode model selection (v1.0.74) in copilot-configuration-basics.md Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-08-01 23:12:29 +00:00 · 2026-07-26 21:04:00 +00:00
parent aa280f28b1
commit c08dd8837a
4 changed files with 31 additions and 7 deletions
@@ -3,7 +3,7 @@ title: 'Automating with Hooks'
 description: 'Learn how to use hooks to automate lifecycle events like formatting, linting, and governance checks during Copilot agent sessions.'
 authors:
  - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-07-13
+lastUpdated: 2026-07-26
 estimatedReadingTime: '8 minutes'
 tags:
  - hooks
@@ -94,7 +94,7 @@ Hooks can trigger on several lifecycle events:
 | `postToolUse` | After a tool **successfully** completes execution | Log results, track usage, format code after edits |
 | `postToolUseFailure` | When a tool call **fails with an error** | Log errors for debugging, send failure alerts, track error patterns |
 | `PermissionRequest` | When the CLI shows a **permission prompt** to the user | Programmatically approve or deny permission requests, enable auto-approval in CI/headless environments |
-| `agentStop` | Main agent finishes responding to a prompt | Run final linters/formatters, validate complete changes |
+| `agentStop` | Main agent finishes responding to a prompt | Run final linters/formatters, validate complete changes; receives `stop_hook_active` flag *(v1.0.72+)* |
 | `preCompact` | Before the agent compacts its context window | Save a snapshot, log compaction event, run summary scripts |
 | `subagentStart` | A subagent is spawned by the main agent | Inject additional context into the subagent's prompt, log subagent launches |
 | `subagentStop` | A subagent completes before returning results | Audit subagent outputs, log subagent activity |
@@ -369,6 +369,17 @@ Run ESLint after the agent finishes responding and block if there are errors:

 If the lint command exits with a non-zero status, the action is blocked.

+> **Preventing infinite loops with `agentStop` *(v1.0.72+)***: If an `agentStop` hook always exits with a non-zero code (blocking every turn), the CLI would previously loop indefinitely. Starting in v1.0.72, the CLI ends the turn after 8 consecutive blocks and passes a `stop_hook_active: true` flag to your hook script in the input JSON, so the hook can detect it is in a forced-continuation scenario and self-limit. Use this flag to break out of blocking logic when appropriate:
+>
+> ```bash
+> #!/usr/bin/env bash
+> input=$(cat)
+> if echo "$input" | jq -e '.stop_hook_active == true' > /dev/null; then
+>   exit 0  # Already in forced continuation — don't block again
+> fi
+> npx eslint . --max-warnings 0
+> ```
+
 ### Security Gating with preToolUse

 Block dangerous commands before they execute. Use the `matcher` field to target only the `bash` tool, so the hook doesn't fire for file edits or other tools:
@@ -3,7 +3,7 @@ title: 'Building Custom Agents'
 description: 'Learn how to create specialized GitHub Copilot agents with custom personas, tool integrations, and domain expertise.'
 authors:
  - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-07-09
+lastUpdated: 2026-07-26
 estimatedReadingTime: '10 minutes'
 tags:
  - agents
@@ -254,12 +254,13 @@ The agent can then query your database, analyze query plans, and suggest optimiz

 | Scenario | Recommended Model |
 |----------|-------------------|
+| Highest-capability tasks, frontier reasoning | Claude Opus 5 *(v1.0.75+)* |
 | Most demanding reasoning, security review | Claude Sonnet 5 *(v1.0.67+)* |
 | Complex reasoning, analysis | Claude Sonnet 4 |
 | Code generation, tool-driven agentic work | GPT-5.6 *(v1.0.70+)* |
 | Code generation, refactoring | GPT-4.1 |
 | Code-specialized tasks, large context | kimi-k2.7-code *(v1.0.68+)* |
-| Quick analysis, simple tasks | Claude Haiku or GPT-4.1-mini |
+| Quick analysis, simple tasks | Claude Haiku, GPT-4.1-mini, or gemini-3.6-flash *(v1.0.74+)* |
 | Large codebase understanding | Models with larger context windows |

 ### Organizing Agents in Your Repository
@@ -3,7 +3,7 @@ title: 'Copilot Configuration Basics'
 description: 'Learn how to configure GitHub Copilot at user, workspace, and repository levels to optimize your AI-assisted development experience.'
 authors:
  - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-07-13
+lastUpdated: 2026-07-26
 estimatedReadingTime: '10 minutes'
 tags:
  - configuration
@@ -468,6 +468,16 @@ The settings dialog supports search — type to filter settings by name. Changes
 /model --local      # view/edit your personal model preference
 ```

+*(v1.0.74+)* Use `/model plan` (or `/model --plan`) to pick a **separate model for plan mode**. This lets you use a lightweight model for planning while using a more capable model for execution:
+
+```
+/model plan              # open the model picker scoped to plan mode
+/model --plan claude-haiku-4  # set a specific model for plan mode by ID
+/model --plan off        # clear the plan-mode model override (reverts to session model)
+```
+
+When you leave plan mode, Copilot automatically reverts to your session model. This is useful for keeping plan-mode fast and cost-efficient while still having a powerful model available for agentic execution.
+
 These flags mirror the **Repo** and **Repo (local)** scope tabs available in the `/settings` dashboard (v1.0.71+), making it easier to manage per-repository vs. user-global configuration without ambiguity. In v1.0.71+, the `/settings` dashboard also shows **Repo** and **Repo (local)** tabs alongside the existing user-level view, giving you a unified place to see which settings are applied at each layer.

 GitHub Copilot CLI has two commands for managing session state, with distinct behaviours:
@@ -3,7 +3,7 @@ title: 'Installing and Using Plugins'
 description: 'Learn how to find, install, and manage plugins that extend GitHub Copilot CLI with reusable agents, skills, hooks, and integrations.'
 authors:
  - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-07-13
+lastUpdated: 2026-07-26
 estimatedReadingTime: '8 minutes'
 tags:
  - plugins
@@ -31,7 +31,7 @@ A plugin bundles one or more of the following components:
 | **Custom Agents** | Specialized AI assistants with tailored expertise | `agents/*.agent.md` |
 | **Skills** | Discrete callable capabilities with bundled resources | `skills/*/SKILL.md` |
 | **Hooks** | Event handlers that intercept agent behavior | `hooks.json` or `hooks/` |
-| **MCP Servers** | Model Context Protocol integrations for external tools | `.mcp.json` or `.github/mcp.json` |
+| **MCP Servers** | Model Context Protocol integrations for external tools | `.mcp.json`, `mcp.json` *(v1.0.74+)*, or `.github/mcp.json` |
 | **LSP Servers** | Language Server Protocol integrations | `lsp.json` or `.github/lsp.json` |
 | **Extensions** | IDE extensions installable via the plugin marketplace (v1.0.62+) | `extensions/` |

@@ -74,6 +74,8 @@ The `plugin.json` manifest declares what the plugin contains:
 }
 ```

+> **Open Plugin Spec v1 *(v1.0.74+)***: GitHub Copilot CLI now supports [Open Plugin Spec v1](https://openpluginspec.org/) plugin manifests alongside the native `plugin.json` format. This means plugins authored for compatible tools using the Open Plugin Spec can be installed directly without reformatting. The CLI also recognises `mcp.json` (without the leading dot) as a valid MCP server configuration file, improving interoperability with other tools that use this convention.
+
 ## Why Use Plugins?

 You might wonder: why not just copy agent files into your project manually? Plugins provide several advantages: