[bot] Add /model --session, /model plan, and copilot plugins install --skill documentation

04-agents-custom-instructions/README.md

@@ -109,7 +109,25 @@ What about the Task Agent? It works behind the scenes to manage and track what i
 
 > 💡 **Multi-turn subagents**: Subagents (background tasks launched by agents) support follow-up messages. While an agent is running in the background, you can open `/tasks` to view it and send follow-up instructions. You don't have to wait for it to finish before guiding it further. Think of it like being able to tap your assistant on the shoulder mid-task to give extra direction.
 
-> 📚 **Official Documentation**: [GitHub Copilot CLI Agents](https://docs.github.com/copilot/how-tos/use-copilot-agents/use-copilot-cli#use-custom-agents)
+### Choosing a Model for Plan Mode
+
+By default, `/plan` uses the same AI model you've selected for your session. You can pick a *different* model to use only while in plan mode — great for using a faster or less expensive model for planning, then switching back to a more powerful one for implementation:
+
+```bash
+copilot
+
+# Open the model picker for plan mode only
+> /model plan
+
+# Or specify a model ID directly (use 'off' to clear the plan-mode model)
+> /model plan gpt-5.6-sol
+
+# Leave plan mode and the model reverts to your session model automatically
+```
+
+> 💡 **Why set a plan-mode model?** A high-quality plan created by a frontier model upfront can actually save tokens and time overall. A precise, well-scoped plan means fewer back-and-forth corrections during implementation.
+
+> 📚 **Official Documentation**: [GitHub Copilot CLI Agents](https://docs.github.com/copilot/how-tos/copilot-cli/use-copilot-cli/invoke-custom-agents)
 
 ---